Subscribe now

Pluribus

The most miserable woman on Earth must save humanity from a virus. Not a virus with the effects one might expect. People don’t get ill or violent. They become connected, peaceful, and completely aligned around a single goal – to spread the virus that brings them together. This is the basic plot of Breaking Bad Creator’s new show Pluribus.

Individuals form a team of billions. The team has perfect cohesion, motivated by the same goal. Each individual can access the skills, technical knowledge and memories of the group. It is the perfect team. They have transcended the limits of a person-bit — a single individual can do more — but as a group they can achieve almost anything: out of many they become one.

It has to start with the individual

Individuals form teams. Teams collaborate with other teams to form organisations. And organisations collaborate, compete and interact. So the performance of individuals has to be the starting point for performance of any group.

I’m not trying to reduce people down to soulless machines, but it’s helpful to think in terms of the mechanics of an individual as a goal seeking machine to uncover what can go wrong with that system:

• An individual comes with certain technical skills and experience

• He/she sets or is given a goal

• He/she moves both subconsciously and consciously towards that goal

• And measures progress towards the goal, feeding back errors (mistakes)

• This has two effects: i) improves capability through learning; ii) allows course correction

Even this simple model helps to illuminate where dysfunction can come from at the level of an individual:

1. The technical skills and experience is completely unmatched to the challenge posed by the environment, so the individual is overwhelmed, cannot progress or learn. This is where the vast majority of attention is usually spent: do we think someone is sufficiently up to the job?

2. An inaccurate assessment of one’s capability vs. the challenge. This is what Patrick Lencione means when he talks about being humble. And it cuts both ways: thinking you are useless is just as damaging as overconfidence, not least because we act outwardly in a way that is consistent with our beliefs and emotions — so believing yourself to be incapable will make it so.

3. Unclear goals or the wrong goals. People will move towards the goals they consciously or subconsciously set. This is why it’s so fundamental to set and clarify goals even for capable people, or maybe especially for capable people — the problem with setting someone capable the wrong goals is that they will achieve them.

4. Progress can be slow and frustrating because of lags between action and effect. The gap between where someone is and where they want to get to generates tension which propels them forward. But we aren’t machines, so as we make that comparison we also experience emotional tension pulling us back, encouraging us to reduce the ambition of our goals. Performance depends on positive tension dominating negative emotional tension, so that you keep driving through adversity and don’t progressively roll back your ambition.

5. The inability to learn from mistakes. Mistakes are inevitable, they are a feature not a bug. It fuels the adjustments needed to keep making progress. So learning has to sit at the heart of performance. There is a challenge here though as it’s not always clear what was a mistake, not least because of complexity and lags between cause and effect. We aren’t always sure of what the lesson is.

6. Agonising over mistakes and not focusing on the goal. High performers learn from past mistakes but they don’t agonise over them. Similar to having an insufficient self image, if one focuses on what has gone wrong previously, it’s more likely to be repeated.

   Subscribe now

From individuals to teams

The requirements we covered above are just as applicable to teams as they are to individuals:

• The team needs the right mix of technical skills and experience to face the combined challenge

• An accurate assessment of where there are gaps and weaknesses, especially as the nature of the challenge evolves

• Clear goals for the team as a whole to drive cohesive action

• Maintaining the creative tension of the team to persevere

• The ability to learn from mistakes, share learnings, and to maintain learnings in the team as people leave

On top of that, there are additional requirements for the group as complexity increases:

• A clear purpose for the team as a whole

• Clear roles within the team

• Coordination mechanisms, rituals and habits

• Clear expectations and accountability for results

• A high level of trust between individuals

• Structures and ways of working that promotes wholeness:

  • Everything is ‘on the table’ and can be discussed

  • Dialogue and the suspension of positions and willingness to test assumptions

  • Tolerance for healthy conflict

Many of these requirements address the need for cohesive action, whilst letting people get on with their jobs and make decisions i.e. balancing the need to act as one whilst providing people with autonomy.

To avoid repeating the sources of dysfunction above, they are all still applicable, but now apply both to individuals and the group. So where does additional dysnfunction creep in at the level of a team?

1. Dysfunction can start from something as fundamental as identity and purpose. Why does the team exist? Does it even feel like a team, or a loose collection of individuals?

2. There’s the basics of everyone knowing what they need to know to sequence and coordinate activities. As an individual it’s straightforward to sequence tasks and understand where the dependencies are. It gets more complicated where projects depend on progress in multiple parts of the org.

3. It can be uncomfortable to hold peers to account so it happens less than it should. Sometimes it’s the result of the things that are left unsaid when making a decision – buy-in that never existed, or a lack of clarity over what decisions were made and where accountability sits.

4. If individuals on a team don’t trust each other – in the sense that they are all there to act in good faith, in a way that is aligned to the team’s purpose, then I don’t see how they could ever be performant. Without it you don’t get healthy conflict, commitment, accountability or results.

5. The Table Group talks a lot about how a lack of healthy conflict is the sign of a toxic culture. I think they rely on that so heavily because it’s a shortcut for knowing that members of a team are being open and honest, and that everyone is contributing. This is a fine heuristic but I’m not convinced that we should be optimising for conflict, it’s just a forcing function to talk openly and then get to a committed decision.

I’ll end this short essay with a quote. Pluribus wasn’t inspired by but does have similarities with the philosophy of Physicist David Bohm’s work on ‘Wholeness’. If we expand the boundary of our identities to include our team members, we are already setting the conditions for a better performing whole.

“Everybody depends on everybody, but actually everybody is everybody”.

David Bohm

Subscribe now

I enjoy writing, but it is a tremendous investment. If you enjoyed this article, please consider sharing it with your colleagues and friends so that it may grow.
Thank you.

Subscribe now

The formula for trust

Transactions depend on trust. We only want to buy from suppliers we feel we can trust. If you believe that, then trust should sit at the heart of how you design and improve the way you go to market. Companies such as eBay, PayPal, and Alibaba, among others, exist explicitly because they provide the safety needed to do business where it was once absent.

I haven’t been able to find who originally came up with the formula below for trust, but I learned it from Patrick Hoverstadt, who ironically learned it from an Anderson Consulting partner just before Enron’s collapse.

Trust = (Credibility * Intimacy) / Risk

Trust is the result of credibility, intimacy, and risk. The greater your credibility in the market, the more people trust you. Assuming you are a credible supplier, the better a prospective customer knows you, the higher the level of trust. Credibility and intimacy are multiplied together, as their effect is double-edged: a credible supplier benefits as a prospective customer gets to know them, while a supplier lacking credibility will be exposed as intimacy increases.

Trust also needs to be put into the context of the risk of the transaction. Low risk, low cost purchases require a lower bar of trust for transactions to complete than high risk, high cost transactions.

Why and how it’s useful

Through this lens, Sales & Marketing is a trust building exercise with prospective customers. The formula reveals specific levers we need to understand, and can then manipulate to drive faster cycle times and increase win rates. We can:

• Increase our credibility in the market, actual, perceived, or both

• Increase our intimacy with prospective customers

• Decrease actual or perceived risk of the transaction

What can we do to improve credibility?

Credibility has two aspects: there is “actual” credibility, and there is what is visible. Strictly speaking, this distinction isn’t explicit in the formula, but it is important. Intimacy uncovers what is true and what is bullshit, but given that prospective customers aren’t going to give every supplier the time of day, signalling credibility to the market becomes as important as having “actual” credibility.

A better formulation would be:

Trust = (Credibility(I) * Intimacy) / Risk

Credibility is a function of Intimacy:

If intimacy is low, then what you see is signaled credibility.
If intimacy is high, then what you see is actual credibility.
Intimacy doesn’t just amplify credibility, it reveals it.

Signalling is something Rory Sutherland talks about passionately in his book Alchemy. For him, presentation matters as much as if not more than the product itself. In fact, the way something is presented directly impacts people’s experience of it (for example, branded painkillers versus non-branded alternatives).

In an era of AI slop, costly signalling becomes more important and more powerful than ever. Which supplier are you going to trust more: the one that spams inboxes with automated SDR slop, or the one that invests in an in-person event? This is where the discourse on AI for GTM goes so wrong: a total obsession with efficiency, while remaining blind to the power of costly signalling.

Regarding “actual” credibility, mistakes generally show up on the right-hand side of the bowtie — customer success, or rather the absence of it. A fantastic product that does not get activated or solve customer problems does little for credibility. It may work for a while, until it doesn’t, inevitably revealing itself through high churn.

Price plays an important role here too. Over the long run, higher prices enable higher margins, greater investment, and increased credibility. When I advised in the sexy world of industrials, it wasn’t unheard of for customers to pay more, knowing that healthy upstream margins help secure their supply chain. As a customer, do you really want your value chain dependent on a supplier that is living hand to mouth?

There is also a common misconception that lower prices automatically lead to higher win rates (the downward-sloping demand curve), but this neglects the role of price in signalling credibility and building trust. If something is so cheap that it feels too good to be true, it probably is.

I have seen prospective customers, usually in enterprise deals, reject suppliers explicitly because they were too cheap, or because they didn’t charge for implementation or service. Low fees, or the absence of fees, signalled low credibility.

Stella Artois is a terrible beer, but it is “reassuringly expensive.” The funny thing is, it wasn’t even that expensive; they simply understood that positioning it as a premium product would help, not hinder, market share growth.

Finally, value selling is an area of growing importance and plays a key role in building credibility and trust during the sales process. It connects real pain to concrete use cases, showing how specific problems will be solved and what impact can be expected.

That said, the single most effective thing I would do as a supplier to build credibility is let current customers sell to prospective customers — nothing beats social proof.

What can we do to improve intimacy?

It’s useful to think about intimacy through the lens of RevOps and the qualification process. Intimacy isn’t a “thing”; it’s a relationship, which means there are two perspectives: sellers and buyers. Qualification happens along the bow tie as a give-and-take of information, helping both parties learn what they need to know about each other in order to proceed.

Well-functioning GTM processes encode both seller and buyer perspectives. The seller needs to ensure the deal is worth pursuing and understand how the buyer will make their decision. The buyer, meanwhile, wants to evaluate seller credibility and ensure there is a high degree of fit.

• What do I need to know about the customer to proceed?

• What does the customer need to know about us to proceed?

• What evidence does the customer need to proceed?

Qualification codifies this into a process, ensuring sellers consistently capture and share the information required. There is a natural sequence to what information is needed first versus later, which also sets the pace. That sequencing helps indicate where the relationship truly stands: are we on a first date, or about to get married? Missed forecasts often stem from a seller’s belief that they’ve achieved a level of intimacy that simply doesn’t exist in the customer’s eyes. The seller has bought the ring, but hasn’t met the parents yet.

The greater the risk involved in a transaction, the greater the intimacy required. More time, steps, and rigour are needed in the qualification process. Buyers will also have their own bureaucracy, processes, and culture designed to protect them and regulate the pace of intimacy. As credibility is demonstrated, intimacy can deepen.

What can we do to decrease risk?

The Jolt Effect is an excellent book for understanding the role risk plays in win rates and how to mitigate it. For those who haven’t read it, the book sets out to explain what top sellers do differently from the rest. One of its key findings is that top performers have a far better appreciation of the role risk plays in the sales process, and they deploy a range of tactics to mitigate that risk for customers.

Mediocre sellers hammer the value side of the equation, emphasising the impact of the solution. They spend their time on the “credibility” variable but forget the denominator. The reality is that even if a seller has done a great job establishing credibility and intimacy, if perceived risk remains high, trust can still be low overall.

The best sellers actively reduce risk. They address customers’ fear of making the wrong decision by making opinionated recommendations, and they reduce the size and scope of transactions when necessary, preferring to close a land deal that they can later expand. Packaging can play an important role here as a startup matures. Typically, startups sell an all-inclusive product to which they progressively add features and functionality. Over time, these propositions become bloated and expensive; packaging then becomes a way to reduce the minimum sellable unit, lower the risk of shelfware, and improve win rates.

Subscribe now

A mental model for design

I believe every seller, marketer, and CEO should have this mental model for trust imprinted in their skulls. In fast-growing startups in particular, leaders are constantly faced with novel situations they need to design for. Having mental models such as the trust formula at hand helps teams quickly diagnose what might be going wrong and focus on high-leverage design adjustments.

I enjoy writing, but it is a tremendous investment. If you enjoyed this article, please consider sharing it with your colleagues and friends so that it may grow.
Thank you.

Share

Subscribe now

What makes a business viable?

I can’t remember how I came across Stafford Beer and his Viable System Model (VSM), probably after reading about something equally esoteric. The model’s power is rivalled only by its obscurity. It is the product of asking “what needs to be true for a system to be viable”, where viability is defined as a system’s ability to survive (and thrive) in a changing environment.

“To succeed you must first survive.”

Warren Buffet

Beer studied the central nervous system to derive a model for viability. Humans have proven they are capable of surviving in a changing and hostile environment over a 300,000 year period. There must be some deep wisdom encoded within us that we can learn from and then abstract to complex systems generally — including for our purposes, business. That is what Beer accomplished.

“Nature has generated by evolution the best control system for complex environments”.

Stafford Beer

The Viable System Model

VSM is a map of how businesses must organise to remain viable — like all maps, it isn’t true in that a map is not the landscape — but it is useful. Now that I have this mental model implanted in my head it’s become much faster and easier for me to diagnose sources of dysfunction in the businesses I support, and identify potential remedies.

The reason VSM isn’t widely known isn’t due to a lack of usefulness: when you read Stafford’s work — or try to interpret his diagrams — your head explodes from the complexity. Unfortunately, he did not make his gift to the world easy to comprehend. It’s a damn shame, because after spending a year trying to get my head around it, I am now convinced of it’s genius. Just as the bow-tie model is now ubiquitous among GTM teams, VSM should be familiar to every manager and leader, of every organisation, no matter the size.

A schematic of VSM — a great way to scare people off investigating further.

When you strip away the complexity, it’s relatively simple. Maybe I couldn’t explain it to a five-year-old, but I think a ten-year-old could grasp the crux of it. Before walking through the actual model, it’s useful to start with what needs to be true for a business to survive over time.

1. Purpose. The purpose of a business is what it does. It will change what it does to ensure its survival. Survival is more important than continuity of purpose.

2. Change. To stay useful, it has to keep an eye on what is happening in the environment. Do customers still want what it’s selling? Based on what it observes and anticipates, it decides how to develop and adapt to ensure its survival.

3. Management. Whatever it does, it needs to do well given its limited resources. It also wants to check that the activities it believes are happening, are in fact happening.

4. Coordination. No business depends only on the actions of one individual, so it must ensure that activities, individuals, and teams are coordinated.

5. Execution. The actual work must get done: writing lines of code, calling prospects, onboarding customers, etc.

6. Information. Finally, all of this can only happen if the right information flows to the right place, at the right level of detail, and with the right frequency.

If we work back from these ‘requirements’, we get the VSM. It is organised into five sub-systems. System One is the lowest level of the hierarchy and System Five is the highest. System One is not the most ‘junior’ and System Five the most ‘senior’ — it is not equivalent to moving up an org chart.

One source of VSM’s power is its applicability at different scales, whether that is at the level of an individual, team, function, business, industry, or even the economy as a whole as Beer attempted with the Chilean economy and his ‘Project Cybersyn’. VSM is recursive because organisations are recursive: individuals sit within teams, which sit within departments, which sit within business lines, and so on.

What is included in each subsystem depends on what you are looking at. If you were to apply VSM to the entire business, the Sales function would be a System One operating unit. You could also treat the Sales function as the system, in which case a BDR team would be an operating unit.

• System One (Operations): the primary operations of the business or department you are looking at. For example, Sales or Engineering functions when applied to a SaaS business. Back office and support functions wouldn’t count as a primary activity as no SaaS business exists to do accounting — finance exists to support the management or development of System One activities.

• System Two (Coordination): how activities stay coordinated, keeping operations within a sensible range by avoiding wild oscillations, resolving conflicts, and standardising operating procedures. This could be accomplished through a mix of people, process and technology. Tools for coordination can and often are simple, for example a shared calendar.

• System Three (Performance Management): manages System One operating units and the here and now. It sets performance expectations, allocates resources, optimises existing processes, and evaluates and rewards performance. It manages the balance between operating units acting as a cohesive whole while maintaining their autonomy. System Three also drops down occasionally to engage with System One operating units directly to: i) verify information it is getting through official reporting channels; and ii) develop a better understanding of what is happening on the ground.

• System Four (Development): where the business looks out at what is changing in the environment, what may happen in the future, and determines how it might need to adapt. Scenario planning, R&D, marketing, and experimentation are all examples of System Four activities.

• System Five (Purpose & Policy): where mission, vision, values, structure, and overall policy is set. It makes the major strategic decisions and manages the balance between preparing for the future (System Four) and delivering on the here and now (System Three).

• Information Flow: across all systems are appropriate flows of information from the environment, across individuals, teams and functions, up towards the Board, and then back down again — delivered through a mix of people, process and technology.

  Subscribe now

Coping with variety

Ashby’s law sits at the heart of VSM: “only variety absorbs variety”. Simply, organisations need to be sufficiently complex to cope with a complex environment, and managers need a sufficient range of responses to match the complexity of the organisation they are trying to control.

Managers, functional leaders, and CEOs are not superhuman — they need to find ways to amplify their own variety, and reduce incoming variety, so that they can cope with the situations beings thrown at them.

“Any color as long as it’s black.”

Henry Ford

Ford relied on reducing incoming variety rather than amplifying the variety of his workforce. His philosophy was to “get it right, and keep it the same’ which allowed him to simplify and optimise operations, adopting a command and control approach to management. It worked because he was able to keep variety between Ford, and the market, and across operations in balance through standardisation. That doesn’t mean the market had low variety, it didn’t, and explains why there was a large aftermarket of businesses supplying various enhancements and customisations to the Model T. Other businesses stepped in to absorb variety of the market that Ford did not want to address.

Minimising incoming variety is still as relevant as it was in the 1920s especially for resourced constrained startups, however, customers are more demanding, competition is more intense, and employees want more than just a paycheck. So there is an avalanche of complexity that needs to be dealt with, and more importantly things just move faster today, so there’s a need to be more adaptable. Internal variety needs to be increased just as much as incoming variety needs to be reduced.

One of the most effective ways to amplify internal variety is to provide autonomy, whether that is at the level of an individual, or operating unit. The people doing the work, who are in the direct flow of information are best placed to make good decisions quickly. If every decision needed to be kicked up the chain of command, progress would grind to a halt. At the same time if everyone were provided maximal autonomy it would obviously be chaotic. VSM acknowledges that organisations need both, there must be space for autonomy, and operating units must also act cohesively through coordinating mechanisms, and through a well functioning System Three.

How to start using VSM

The immediate benefits of VSM without needing to be an expert in it (in my opinion) is to use it as a diagnostic tool to quickly identify sources of dysfunction and gaps in capability. I’ve listed out a number of key questions that I now use when working with Notion’s portfolio.

• Are individuals clear on what their job actually is, and what activities are expected of them every day, week, and month?

• Are activities and teams coordinated?

• Do we have the right metrics and visibility across teams to manage performance in the here and now?

• Have the targets we’ve set reflect the resources we have equipped operating units with?

• Do we have the information (both internal and external) we need to make the decisions we want to make?

• Are managers conducting skip level meetings, and diving deeper into the business to validate reporting and learn from the ‘factory floor’?

• Have we got the right balance of focus and investment between managing the here and now vs. developing for the future?

• Do we have a strong purpose which orientates the business, is well understood across the business, and guides decision making in the absence of explicit instructions?

This has only been a short intro to VSM which has significant depth not captured here — I hope though it has inspired you to dig deeper and unlock the benefits for your business.

I enjoy writing, but it is a tremendous investment. If you enjoyed this article, please consider sharing it with your colleagues and friends so that it may grow.
Thank you.

Subscribe now

Subscribe now

How to think about fit

The transition between the initial success of founder-led sales and a repeatable engine of growth is when many start-ups get stuck. The potential from having achieved a degree of product-market fit does not translate into revenue. So much promise. Wasted.

The gap looks small but it’s a chasm – much harder to cross than it seems. Founders and their teams need to build the machine that will deliver growth, whilst hitting ambitious revenue targets quarter-on-quarter. They fly the plane whilst trying to build it. ‘Fitness’: the relationship between the business and the market is in large part what will drive the revenue outcomes they are looking for. A way to visualise what’s going on is to use a fitness landscape. Every point on the landscape is a particular strategy and configuration of the business, where height represents outcomes.

A small adjustment to coordinates on the landscape can equate to a large change in height. Fitness is sensitive to small changes in your go-to-market strategy, whether that is adjusting your ICP, messaging, resourcing, and so on. If you knew what the landscape looked like, you could design your strategy and configure the business in a way that would maximize fit. But you don’t. Building up an understanding of your customers, what they want and how they buy, helps to bring some clarity – but decisions are still made under uncertainty. The path forward is rarely obvious.

Good mistakes, and bad mistakes

Strategy isn’t top-down or bottom-up; it’s both. Top-down decisions primarily drive focus and knock out what not to do. It sets boundaries around where on the landscape the business should play. This can also be informed bottom-up. As more of the landscape is explored, more of the landscape can be ruled out.

This bottom-up search process isn’t sophisticated and it doesn’t need to be. The reality is that teams use simple heuristics to make progress – they try new things, stop doing the things that don’t work, and do more of what does. Thousands of small steps forward. The path to go-to-market fit is one of continuous learning and refinement. Taking steps in the wrong direction is part of the process – your path to go-to-market fit has to have a high tolerance for errors. It’s what fuels the learning process. Making mistakes is a feature, not a bug.

At the same time, Series A startups are, in particular, severely resource constrained and may only have 18 months of runway before they need to raise more capital. They can’t afford to make more than one or two big mistakes. This seems at odds with ‘mistakes that are needed to fuel learning’. What matters are the types of ‘mistakes’. Low consequence errors that teach you something are helpful. Betting the farm on what ends up being the wrong direction is not. Bottom-up experimentation provides a lot of the insight needed to make those bigger top-down bets. Accountability for those bigger bets sits with founders to get right, whether they have expertise in the topic area or not.

More complexity than you could ever handle

I believe there are three high-level dimensions to go-to-market fit that founders and their teams need to get right.

1. Orientation: where in the market they will focus their efforts.

2. Systems: how they will find demand and translate it into paying customers.

3. Resources: how they will allocate resources (people and capital).

Orientation requires decisions over how to segment the market and then which segments to pursue. Do you segment by industry, size, use case, region, etc.? Then, which of those segments will you orient the business around, which companies, departments and personas will you target. At the same time you are building the machine – making decisions over channels, qualification frameworks, sales and marketing technology stack, and so on. Finally you deploy headcount and capital in the right places.

Below is a relatively simple option space that shows the number of options that could form the basis of a go-to-market strategy. For example: six ways to segment the market, four channels to market, etc. The size of the option space – equivalent to coordinates on our fitness landscape – is calculated by multiplying the different options across variables. The example below equates to 1.8M coordinates. 1.8M different ways to configure the business. That should be sobering. That’s the level of complexity a business navigating to go-to-market fit needs to contend with.

For all practical purposes, the chances that a team gets lucky and designs the optimal strategy are zero. Wandering the landscape, making small adjustments, isn't going to allow teams to learn and adjust fast enough before running out of runway. How is it possible that any business, facing these sorts of odds, gets it right?

Cut complexity in half, again and again

We need a way to radically reduce the 1.8m possibilities to have any chance of making progress. Every decision can be restated as a go/no-go decision. This is represented by taking the log base 2 of each variable and summing the results. In this case 22. The magic of this approach is that it reduces an option space of 1.8m coordinates to just 22 go/no-go decisions. That starts to sound much more doable.

22 decisions represent the maximum complexity we need to manage. In reality, many of these decisions will be interdependent. For example, choosing to focus on a particular segment might eliminate several personas that aren’t relevant. Mapping out interdependencies reduces the number of go-to-market designs that make sense in reality. For example, you aren’t going to hire enterprise account executives if your acquisition motion is product-led. That would be insane.

A massive reduction in complexity is driven by your choice of customers. That decision has a bunch of knock-on effects on the remaining decisions. Not committing to a section of the market to fight for is so damaging because it retains enormous complexity. Not only are you failing to reduce the option space, but you will also need to build more internal capabilities to satisfy very different fitness conditions between segments. For example, if you choose to serve large enterprises and a prosumer segment, you are going to need to build two completely different sets of capabilities.

Where does this leave us? Getting to go-to-market fit, developing a strategy in general, involves navigating complexity under uncertainty. Significantly more complexity than is generally recognised as I demonstrated with the option space above. To have any chance of succeeding founders and their teams need to:

• Narrow down their ICP as much as possible to drive focus;

• Map out the consequential decisions they need to make and their interdependencies;

• De-risk those decisions through small scale testing before making large costly bets;

• Keep the main thing the main thing - avoid distractions (which have an outsized impact on complexity).

I enjoy writing, but it is a tremendous investment. If you enjoyed this article, please consider sharing it with your colleagues and friends so that it may grow.
Thank you.

Subscribe now

Subscribe now

It’s impossible to know which startups will become blow-out successes — if it were easy, most investments wouldn’t lose money or yield small returns. Out of 20 investments, only 2 or 3 will reach their potential. The rest, despite starting with just as much promise, get filtered out along the path to $100M in revenue. Venture capital invests in the world’s optimists, founders who have total conviction in achieving a highly improbable outcome. A lot of time is spent trying to identify what separates the ones that make it from the rest - the differences that make a difference. Time should undoubtedly be on that list.

“The speed at which a company learns, iterates, and executes is the strongest predictor of its success. Executing with urgency always beats a perfect plan. Amazon, Tesla, Apple, and Nvidia share one critical trait they’ve maintained despite their enormous scale: the relentless ability to reinvent their businesses and capitalize on emerging opportunities faster than anyone else in their respective markets.”
Kamil Mieczakowski, Notion Capital

Why it matters - but does it always matter?

First of all, venture capital funded businesses are only viable for as long as they have external funding. Funding buys time to make the progress needed to secure the next round. Insufficient progress creates a distressed situation for even promising businesses. Less funding, or a premature shift towards break-even can quickly knock a fragile startup off of the venture track.

“Move faster. Slowness anywhere justifies slowness everywhere. Moving fast compounds so much more than people realize.”
Sam Altman

Secondly, as we are seeing with AI, the environment is moving much faster than it was even a few years ago. New announcements come thick and fast. The latest change as I write this is the release of DeepSeek. Something OpenAI and others will need to make sense of. The faster the external environment changes, the faster a business needs to move. If it does not match the speed of change on average it will eventually lose fit and decline.

Matching rates of change also applies to internal functions. Their tempo and rhythm needs to match if it wants to act as a cohesive whole. For example, Sales' ability to acquire new logos faster than CS can onboard them does not lead to better performance of the system as a whole - and probably a lot of internal and customer friction. Timing matters - not just speed. Doing something out of sequence, no matter how quickly, is in my opinion more damaging than moving slowly - it wastes valuable funding as well as time.

Subscribe now

I feel the need, the need for speed

The speed of what? How fast you build a product, or acquire new customers usually comes to mind. But that’s not all. In Patterns of Strategy, Patrick Hoverstadt covers three aspects to time:

• Cycle time: how quickly you can make decisions and operate a core process

• Foresight: your ability to look ahead and anticipate possible futures

• Change rate: how quickly you can adapt existing capabilities and build new ones

Given the low maturity of startups, and their fragile state, the speed at which they can make decisions and learn is particularly important. John Boyd created the OODA loop specifically to speed up decision making. Developed for fighter pilots, the idea was to cycle through the loop faster than your opponent can, influencing and disrupting their actions. In a competitive game, you want to “get inside your opponent's loop” so that they are reacting to your tempo rather than setting it. An OODA loop is used by emergency drivers in the UK, embodied in the system of car control that enables faster and safer driving. In this case there is no opponent. It isn’t about disrupting a competitor. So OODA is applicable to single player and adversarial activities alike.

OODA stands for observe; orientate; decide; and act. One could look at each phase in turn and find opportunities to speed things up.

Observe: How fast can we collect information?

Whether we are talking about a fighter pilot, emergency driver, or startup, observation happens across the entire loop. Information fuels decision making. Having the right information available is a source of speed. So is giving decision making authority to the people who have the information to act. Speed is one of the core arguments for distributed authority.

Internal information requirements should be straight forward to get a handle on. It just takes some effort and investment. RevOps plays a critical role here ensuring an appropriate data model and reporting is set up. Unfortunately there is a general lack of investment in data infrastructure and reporting no matter the size or maturity of business. Informal information flows matter just as much as formal, sometimes more. A big loss of working from home is missing out on these water cooler conversations. There is a good argument to be made to move people around an organisation so that their expertise and knowhow can spread more effectively through the business.

Informal and formal networks that bring information from outside act as valuable invisible assets. Having formal structures set up such as customer advisory boards, regular customer surveys, or QBRs saves enormous time. Looking back on my time as a consultant where a client would spend $100k+ and months to conduct customer research seems absurd. They should have already been plugged into their customers through established channels to ask the questions they need to quickly.

Informal relationships that stretch outside of the business are just as valuable as the internal water cooler moments. Relationships between functional peers through e.g. the Pavilion network, or investor connections are low cost and can be extremely valuable. A key benefit of being part of Notion Capital’s portfolio is the formal and informal networks we set up for our investments. Functional events, WhatsApp groups, webinars etc. all exist to get people across the portfolio connected and talking.

Orientate: How fast can we incorporate new information?

At the heart of the orientate phase is information processing and modelling. An established model allows information to be processed quickly and the implications ‘spat’ out for decision makers to consider. If you lack a model you need to build one, that takes time. For example, market entry decisions are quicker if you already have a framework in place. You don’t need to re-build an approach from scratch every time you consider entering a new market. Models don’t have the be formal and explicit, they could be based on simple heuristics or implicit based on cultural norms and accepted organisational insight.

Decision: How fast can you decide on a course of action?

If the right information is available, and a suitable model is in place to assess the implications of different routes forward then decision making should be relatively quick. Splitting decisions into one-way (irreversible) vs. two-way decisions speeds the process up further for lower consequence decisions. It protects against over-analysis of decisions where it would be better to act, review and adjust than agonize over what to do. Decisions can still get stuck, especially when a functional leader is trying to optimise for their function first, rather than the business as a whole - in those cases it’s necessary to disagree and commit.

Act: How fast can we act?

Action will depend on if you have the capability and resources in place to execute, and to what extent existing capabilities have been optimised. If the right capabilities to execute aren’t in place, speed depends on Patrick Hoverstadt’s ‘change rate’ dimension. There is a tension here between optimised capabilities, and retaining flexibility to make fundamental changes or build new capabilities.

Highly optimised processes have faster cycle times. By definition there is less redundancy and flexibility in the system. Founders face a dilemma: optimise and drive greater efficiency, or maintain a level of redundancy to pivot when needed. If you compress time for a given process, you are elongating time when it comes to making a meaningful change to that process. The insight here is to be thoughtful over the capabilities you choose to optimise. You may be gaining time in the short term and degrading your ability to make changes in future.

Feedforward: What might happen?

Feedforward isn’t a phase. It happens along the entire loop. As information comes in, you feed it forward through your models to anticipate the action you may need to take. It can be used to update models before decision points are reached and further speeds up decision making. Shell uses a version of feedforward, scenario planning to speed up decision making. They have been using scenarios as a key input to their planning process for over 50 years. Its purpose is not to predict the future, but to model possible futures. The process helps managers to confront potential blind spots and make quicker decisions.

Focus

“Our companies ran at higher velocity, with higher standards and a narrower focus than most. Going faster, maintaining higher standards and with a narrower aperture.” Frank Slootman

Slootman’s businesses didn’t run faster with a narrower focus. They ran faster because they had a narrower focus. Focus allows concentration of resources. Think about the enormous energy required for SpaceX’s Starship to lift off. 33 boosters pointing in the same direction. None of those 33 were pointing sideways or in random directions. It’s common sense - if you have resources concentrated on a narrow goal - you are going to move quickly.

Were you rushing or were you dragging?

Terence Fletcher: Why do you suppose I just hurled a chair at your head, Nieman
Andrew: I-I don't know.
Terence: Fletcher: Sure, you do.
Andrew: The tempo?
Terence Fletcher: Were you rushing or were you dragging? Andrew: I-I don't know.
Whiplash

A more nuanced view of tempo than ‘faster is better’ is to move at an appropriate speed depending on the circumstances - not rushing or dragging. Rushing could translate into a lack of focus, executing out of sequence or an accelerating burn; dragging, a lack of progress and missed opportunities.

Slow is smooth, and smooth is fast - a saying originating in the US Navy reflects the sentiment that faster isn’t always better. In a military context it refers to building up accurate muscle memory during training, and then executing movements precisely in the field under pressure. Moving slowly, to build up muscle memory buys time later. One of the biggest fuckups we see is when startups rush for growth when they haven’t built up muscle memory - expanding Sales headcount despite a broad ICP, uninspiring messaging, and a lack of a repeatable sales motion. “More companies die of indigestion than starvation” - trying to run before they can walk.

The reality is that every business is in a perpetual state of under- and over-investment. It’s inevitable given the complex web of dependencies between functions, and lags between making an investment and seeing its impact. A sensible heuristic is to focus on identifying and removing key bottlenecks, combined with an appreciation for the lags inherent in the system. The best founders use foresight to remove limiting factors in the right order, so the business builds and maintains its momentum.

I enjoy writing, but it is a tremendous investment. If you enjoyed this article, please consider sharing it with your colleagues and friends so that it may grow.
Thank you.

Subscribe now

Subscribe now

The crux

If the anticipated productivity shock from AI mirrors the scale of the Technological Revolution it will create deflationary pressure for entire economies, and cut prices at the knees in multiple industries. We will see much less dramatic shifts in pricing models - predictions of businesses embracing outcome-based pricing in particular are overblown.

The macro view

New technologies and the productivity we enjoy from them are built on the shoulders of giants. The AI revolution will build on the innovations that came before it. Technological progress and capital deepening i.e. augmenting labour with capital investments have played a major role in driving growth over the past 200 years.

Ln of GDP per capita in GBR 1700 to 2022

The Technological Revolution is a particularly interesting period to study. New manufacturing processes, increased electrification, adoption of telegraph networks and railroads, and greater coverage of basic utilities all played a part in driving massive increases in productivity. The productivity shock was so consequential that between 1870 and 1890 prices dropped by 0.40% per year on average. Some years saw decrease by as much as 4%.

GBR annual inflation during the Technological Revolution

“History doesn’t repeat itself, but it often rhymes” 

Mark Twain

The nature of innovations ahead of us will obviously be unrecognisable to what emerged in the Technological Revolution. But if the depth and breadth of the coming productivity shock is comparable, we might see similar pressure on prices.

Despite expected productivity gains from AI, the overall picture for inflation is uncertain. We don’t know how monetary policy will develop, what a Trump Presidency will look like, in particular regarding trade. Additionally, energy starvation, overregulation and supply chain disruptions could all threaten the pace of AI adoption, meaning that productivity gains don’t materialise for some time.

If the UK and Germany continue to commit slow national suicide through their respective energy policies, and the EU continues to overregulate then that’s bad news for Europe. If China blockades or invades Taiwan before TSMC can build up production capacity in the US and elsewhere it will be catastrophic for AI supply chains.

Energy starvation (source: Financial Times / BloombergNEF)

National suicide (source: Deutsche Bundesbank; Boom and Bust report)

Supply chain fragility (source: Counterpoint Foundry Revenues Tracker)

Putting aside any headwinds, AI adoption will flow through the economy as follows. Augmenting labour with AI will make workers more productive and increase growth rates. Increased output driven by productivity gains will allow the economy to increase savings, investment, and capital per unit of labour. Economies will move towards higher output, savings and investment as they realise the full potential of AI.

AI as a positive shock to productivity…

…which drives higher output and investment

AI related investment as % of GDP (source: Goldman Sachs)

As businesses become more productive, increased output will put downward pressure on prices. Either we will see deflation, or disinflation.

Increase in aggregate supply…

…puts downward pressure on price levels

Share

The micro view

The irony is that the sectors and use cases which experience the greatest productivity gains are also the ones that will come under the most price pressure. 41% of code on Github is supposedly AI generated. If you can write more code with the same number of engineers, or hire fewer engineers for the same output, those lower costs will flow through into market prices. AI won’t just increase productivity in high tech industries - it will also have a profound effect on under-digitised yet essential industries. The same industries that drove productivity gains during the Technological Revolution.

Cogna, a precision software factory powered by AI is building applications for under-digitised industries such as utilities and construction. It can do so at 1/20th of the cost of software development consultancies. This isn’t fantasy, it’s happening right now. Not only will that put enormous pressure on price levels in the $363bn IT consultancy market, it will also open up new markets as the cost of custom applications becomes more accessible. Increased digitisation and productivity improvements in these vital industries will flow through to lower prices for consumers and businesses. The ability of AI to help citizens keep more money in their pockets through lower prices of life’s essentials is what in large part drives the founders of Cogna.

“Software has long been constrained by a shortage of engineers, leading to broad, configurable products that often compromise usability or require costly customisation. While demand for custom software exists, high costs have kept it inaccessible to most businesses, despite agile methods showing the value of iterative development.

AI is revolutionising this landscape by drastically reducing discovery, development, and iteration costs. This shift enables new use cases and leaner, more targeted solutions. At Cogna, we believe AI-powered, precision-built software can unlock unprecedented productivity, reduce costs, and enhance accessibility for businesses and consumers.”

Ben Peters, Co-founder & CEO of Cogna

AI’s impact will therefore be one of creative destruction. Any excess margins in the short term will be competed away as the underlying technology driving the revolution is available to everyone. Suppliers that grow margins through increased productivity will only keep hold of the gain if they posses market power. Amazingly, even a supplier with market power will reduce prices to increase volume in response to an increase in productivity. It’s profit maximising point will be a function of higher volumes at lower prices enabled by a lower cost base. However you look at it, prices are coming down.

Subscribe now

Market power and monetisation

How you charge has nothing to do with developing market power. Your monetisation strategy determines how much value you capture, not how much you create. To argue otherwise is to get the relationship backwards.

I care just as much about the things that stay the same than the things that will change. Principles that survive massive change are robust. We can trust them. The ways in which businesses generate market power is going to be completely unaffected by AI. I have great faith that Hamilton Helmer has already nailed it by identifying seven sources of power. AI will not make any of them more or less true, and does not suddenly open us up to a new eighth power. I’m not even sure any of them will become more or less important overall. We will probably see more counter-positioning as AI disrupts existing business models, but that will peter out. I’ve summarised Hamilton’s seven powers below if you are not familiar.

Developing market power is what secures profitability and market share in the long term. AI driven innovations will provide a window to do so, but won’t be enough on their own.

“People who invent a great new product add to the “giants shoulders” effect and that influences future inventions but usually not in a way that allows the full stream of profits from these future inventions to flow back to those who came up with the earlier inventions.”

Karl Whelan, Professor of Economics at University College Dublin

What you charge > how you charge

My former employer, Simon-Kucher & Partners, global leaders in monetisation have a saying: how you charge is more important than what you charge. But as I’ve laid out above, the creative destruction we will see as a result of AI will be driven by deflation not creative pricing models. Predictions of large scale shifts towards outcome-based pricing in particular are misplaced.

Pricing according to outcomes is nothing new. Businesses would love to price based on the true value they deliver: capturing a higher share for themselves. We don’t see it much because it’s hard to do. Artificial intelligence doesn’t make it any easier. So no, we aren’t going to suddenly see a huge shift towards outcome-based pricing.

“Outcome-based pricing isn’t new, despite hype suggesting groundbreaking innovation. Recruitment, real estate, and consignment sales have used it for years. However, it’s a poor fit for most things - especially in digital contexts and where event volumes are high, but values low. The fraud-plagued digital advertising market is a good example, where unclear attribution models created chaos. Emerging technologies like GenAI or "Agentic AI" don’t make this model any more viable. Unless you can clearly link costs, autonomous delivery, attributable outcomes, and value it’s best to steer clear for now.”

Marek Rubasinski, Head of EMEA m3ter

What we will see is a shift away from user-based pricing for propositions with a high degree of automation disconnected from labour intensity. However, it won’t disappear completely despite grand predictions on LinkedIn - it still makes sense for propositions that augment a worker’s productivity, blending man and machine. The argument that augmenting labour with AI drives a reduction in labour, so per user models aren’t suitable is erroneous. Labour intensity decreases and capital increases because of the productivity shock. Changing the model doesn’t mean more value is going to necessarily accrue to the supplier - market power matters much more than than the model.

The other argument I see for claiming per user or fixed subscription pricing is dead is that LLMs can drive material variable costs for the application layer. That is true but ignores dynamics. Variables costs will continue to come down as the technology matures. Look at what happened to mobile phone contracts over time. We saw a shift from pay-as-you-go to all you can eat per SIM contracts as investment increased, technology matured, and competition intensified. Same for broadband.

There is already a general shift towards usage-based and hybrid pricing models over the past few years, pre-dating the emergence of LLMs. According to OpenView’s “The State of Usage Based Pricing”, 61% of SaaS businesses already have some kind of usage element, and another 21% will experiment with it. AI will accelerate the adoption of hybrid models but did not start it. Counterintuitively, Geoffrey Moore argues we may see more fixed-fee models devoid of any creativity as AI eats into low-value billable hour services. Increased sophistication and flexibility in the quote-to-cash process is going to be a much more important driver of creative pricing in future than AI.

The desire for more imaginative pricing has always existed, but the Q2C tech stack has acted as an insurmountable barrier holding back a wave of creativity. That is why Notion Capital invested in m3ter, which automates complex bill calculation through a metering and rating engine. m3ter provides the missing puzzle pieces for suppliers to execute the pricing they believe is right, not just what their Q2C process allows.

I enjoy writing, but it is a tremendous investment. If you enjoyed this article, please consider sharing it with your colleagues and friends so that it may grow. Thank you.

Share

This month’s article dives into complexity science, covering a selection of the most important systems laws and principles to help develop strategy and drive change.

Subscribe now

Why this is important

We are allergic to playbooks at Notion Capital. A playbook mindset stops people from the difficult job of thinking. A tactic or strategy that worked in one business is likely to fail when context changes. Understanding why something worked, or didn’t, allows us to make adjustments as circumstances change. Principles over playbooks.

Businesses are complex systems. If we want to understand the principles and laws that govern them, we should be studying complexity science for answers. Look in places others don’t, and we are rewarded with a novel and powerful toolkit to develop strategy and drive change.

Managing complex systems

Businesses are complex, not just complicated. Complex because they consist of many moving parts; operate under uncertainty; in an environment that is constantly changing. VCs and startups talk about growth loops and flywheels - examples of closed loop systems. Closed loop because there is feedback, where outputs become inputs.

The basic components include:

• some goals

• a controller, for example a manager

• the system in question e.g. an SDR team

• output/performance of the system

• measurement of variances

• feedback to close the loop

How you draw a boundary around the system you are interested in depends on what you are trying to achieve. We can draw a box around the diagram above and call it a system. Every system sits within another system. They are recursive, and have a hierarchy. For example, an SDR team sits within the go-to-market function, within a business, industry, and economic system. Each is a system in its own right.

We hear a lot about the role of people, process, and technology, but nothing about the laws and principles that govern the systems we are trying to manage. It’s like focusing exclusively on the driver, engineers, and car of an F1 team without ever mentioning aerodynamics or grip. Cybernetics is the study of these laws and principles - a branch of complexity science applied to management.

“If cybernetics is the science of control, management is the profession of control.”

Stafford Beer

I am indebted to the systems thinker and author Patrick Hoverstadt for much of the framing and thinking I’ve laid out below, and would encourage anyone who stumbles on this article to seek out his work. I adapted the below from his framing of management approaches in particular.

Paradigm one: Garbage can management

Managers walk around with two garbage cans. One contains problems, the other solutions. When they come across a problem, they dig into their can and pull out a solution. If things are slow, they will navigate the business with a solution in hand, looking for a matching problem. They don’t operate with models or frameworks in mind, and are totally blind to second order effects. Instead they focus on immediate symptoms and fixes. Infantile management.

Paradigm two: Playbook reflex

Managers operate with models or frameworks, but they deploy them from ‘off the shelf’. They don’t adapt their approach based on context, expecting to achieve similar results in new situations. A mindset that is especially noticeable in managers that have achieved some success. Blind to the differences between contexts that really matter, they enter new situations with a deadly mix of hubris and incompetence.

Paradigm three: Modeling

Managers operate from a principles over playbook mindset (even when they talk about playbooks). Building or adjusting models to suit the problem and context they are looking at. They close the loop - adjusting their models based on real world feedback. Embracing experimentation, testing assumptions and refining their models to maintain fit with the environment.

Subscribe now

Boundaries & perspective

Before we can manage a system we need to first define it. We do this by drawing a boundary around what is included and excluded. If we have a churn problem, do we draw a boundary around Customer Success only, or do we include Product? What about Sales or even customers themselves. Where we set those boundaries changes the system as well as our perspective, analysis, and conclusions.

Making a distinction

"We take as given the idea of distinction and the idea of indication, and that we cannot make an indication without drawing a distinction."

The quote above is from George Spencer-Brown’s Laws of Form. He’s pointing out that to talk about something you have to first make it distinct from everything else. It’s the fundamental starting point to systems thinking.

By drawing a boundary we are implicitly making a statement about what matters. We are stating that the thing we have drawn a boundary around is something distinct. Customer segmentation is an example of boundary setting. We carve up the market in a way that highlights the distinctions between groups of customers that really matter. Defining your ICP is a critical boundary setting exercise to help minimise complexity.

Perspective

“It is a narrow mind which cannot look at a subject from various points of view.”

The boundaries we draw, and where we sit in relation to them shapes our perspective. Which in turn drives the decisions we make. For example, does a CMO draw a boundary around the marketing function she leads, or around the executive team she is a member of, or the revenue function as a whole? Each boundary creates a different perspective. If she primarily sees herself as the leader of marketing, she will make different decisions vs. as a member of the executive team. Both perspectives matter, but her ability to cross boundaries and switch between perspectives will shape her models, objectives and decisions.

Different perspectives of the same function. None are wrong, but none capture the full complexity either.

Our perspective changes whether we sit outside, at the edge of, or well within a boundary. For example, customer success managers often sit right at the edge of their company's boundary. From there they can empathise with customers, providing invaluable insight to product management. Straying too far into the customer’s world though creates issues. They become their customers’ best friend and lose the ability to have difficult conversations.

Finally, we need to be aware of how boundaries create communication challenges. What makes sense within one boundary, might not make any sense when the message crosses into another system. Comms is such an important function because messages crossing a boundary from one system to another need to be translated so that they still make sense. For example, the decisions made at Board level, and the rationale behind those decisions require significant translation as they move across multiple boundaries. Not only vertically in terms of hierarchy but also horizontally across functions.

Share

Decision making & control

Effective management relies on making good decisions quickly, translating them into action, and adjusting based on feedback. Modeling sits at the heart of this.

Contant-Ashby theorem

“Every good regulator of a system must be a model of that system.”

In relation to management, if you want to be a good manager, you need to have a model of the thing you are trying to manage. It doesn’t need to be an exact reflection of reality, but must capture the structure and dynamics that matter. Having a model in place allows you to orientate and speed up decision making. If you then plug real world feedback into the model, you can adjust and improve it over time. The rise of RevOps is a clear sign of the growing recognition of modeling and its impact on performance - practical systems thinking at its best.

The London tube map is a useful model if you want to navigate the underground. Useless if you want to walk it. You’ll quickly find the distances between stations are distorted, and the different lines don’t give you much guidance over where you need to go. So the models we build need to be fit for purpose - recognising that all models are wrong, but some are useful.

Law of requisite variety (Ashby’s law)

“Only variety absorbs variety.”

To be effective, a manager must have at least as many ways to react as the variety of situations he might face. You can increase your variety of control, or you can reduce incoming variety. What matters is that they match. Reducing incoming variety is especially important in the early phases of a start-up as so little process and policies are in place. Over time, it can build more internal variety, which allows it to take on more complexity from the environment e.g. serve additional segments of the market.

In competitive situations, the player or team with more variety will tend to win. Greece unexpectedly won the UEFA European Championship in 2004 because they deployed a variety of play that opposing teams couldn’t match. They changed formations depending on who they played, had a formidable defence that was difficult to break down, well drilled set piece plays, and a team that worked seamlessly together. A much less talented team beat the greats of Europe. That’s not luck, it’s Ashby’s law at work.

Redundancy of potential command principle

“Power resides where information resides.”

The first implication is that as long as you can get the right information to decision makers, you can distribute authority. Distributing information across the organisation becomes synonymous with distributing authority. Information becomes the key ingredient to move away from a command and control approach.

Second, decision makers will make the decisions that they have the information available to make. So if you want quick decision making, information needs to be prepared in advance. Collecting information before you need it makes it technically redundant, which leads us to our final implication.

Third, having spare / redundant capacity is a source of resilience. If one part of the business goes down, another can step in. This is only possible if information is distributed.

Balancing autonomy and cohesion

“Managers should maximise the freedom of colleagues, within the constraint of fulfilling the business’s purpose.”

This touches on Paul Graham’s flawed essay on ‘Founder Mode’. The key point he should have been addressing is the tension between providing colleagues with autonomy, whilst ensuring that everyone is working together in a cohesive way. This has nothing to do with professional managers vs. founders.

Maximising the freedom of colleagues within constraints has both moral and practical dimensions to it. Moral, because everyone wants to live a life free from control and coercion. Practical, because autonomy gets you speed, flexibility, and the capacity to deal with massive complexity. The right balance depends on the nature of the challenge and the purpose of the system. 

An F1 pit crew needs maximum cohesion, and minimal autonomy during a race. Everyone has to be in exactly the right place, at the right time, executing in perfect synchrony. Imagine the chaos if each member had the autonomy to decide what to do and when. On the other end of the spectrum is an improv comedy show. Each comedian has a lot of freedom to express themselves. The need for cohesion is minimal. They just have to be on time, and act in a way that’s consistent with their purpose - to make people laugh.

Recursion is at the crux of why this balance is universal to absolutely every team and colleague. You can always move up or down the hierarchy and find functions which want cohesion from the teams and individuals that sit within the function, and autonomy from the hierarchy above.

Subscribe now

Change & survival

Performance is systemic. It’s an emergent property of the whole business and its relationship with the market. As the market changes, so to must the business in order to maintain fit. Otherwise it wont survive.

POSIWID

“The purpose of a system is what it does”

Performance is a perfect reflection of the business you’ve built. If you want to see different outcomes, you have to change its structure or orientation. A massive red flag is when a business isn’t growing and the CRO makes statements to the effect of “failure isn’t an option, we have no choice but to grow”. No one doubts intentions - but trying harder isn’t a strategy.

Gall's Law

"A complex system that works is invariably found to have evolved from a simple system that worked. A complex system designed from scratch never works and cannot be patched up to make it work."

Gall’s law highlights the dangers of over-engineered solutions. Iteration provides the time needed to test and learn before committing. Simple unoptimised solutions provide more flexibility to adjust as circumstances change. Simple is also cheaper. You can keep burn low by matching complexity to the scale of the challenge, focusing instead on doing just enough to remove limiting factors.

It explains the popularity and effectiveness of the ‘Lean Startup’ approach, in particular building an MVP which can then be used to test and learn. The same logic applies to new processes. Starting with minimum viable processes, and then adjusting through real world experience, rather than over-engineered approaches which end up being misspecified. 

Fitness

“Survival of the fittest.”

Fit is one of the few principles that is talked about extensively, but it's applied too narrowly to a product and its market. Instead, it is a principle of universal applicability. Relevant to the relationship between a business and industry or founder and evolving start-up. The implication for founders is to change and evolve as their business grows, maintaining fit with what the business demands of them. And for the business, to match the average rate of change of its industry to maintain fit.

Information plays a central role to building and maintaining fit, which I’ve written about here and here. Hiroyuki Itami covers this elegantly in his book ‘Mobilising Invisible Assets’. The logic works as follows: accumulating information drives fit, which drives growth, which requires a series of overextensions. No different from an individual pushing themselves outside of their comfort zone to keep learning.

Closing remarks

The cross-discipline nature of complexity science is what makes it so powerful and interesting. Researchers from disciplines who historically did not work together, or understand each other have been collaborating increasingly closely over the last 70 years, creating a new kind of science. They were the weirdos, and outcasts, whose work is now slowly going mainstream. There is so much the business world can learn from this new science, bringing a level of rigor it badly needs.

“The twenty-first century will be the century of complexity.”

Stephen Hawking

I enjoy writing, but it is a tremendous investment. If you enjoyed this article, please consider sharing it with your colleagues and friends so that it may grow. Thank you.

Share

This month’s article looks at what makes a business intelligent; it’s structure and properties. A short primer on a complex topic.

Subscribe now

What is intelligence?

Writing an article about what makes a business intelligent is tricky when there is no agreed definition. And no, IQ doesn’t count. Fortunately, there are a lot of intelligent people (whatever that means) studying the topic that we can learn from. Understand its properties then maybe we can operate more intelligently.

Competition is trying to steal your customers. New technologies threaten you with obsolescence. Governments change the rules of the game. Surviving and growing in hostile markets depends on your ability to take in information, select the best response from your range of possible actions, and adapt as the market changes.

On information and variety

You take the decisions that you have the information to take, and power is where information in the business resides. This is the “Redundancy of Potential Command Principle”. Timely inputs, in the right mix, and enough of it is needed to drive high quality decisions. To go beyond centralised command and control, information must flow. Vertical flows to maintain control; horizontally to ensure decision making centers are coordinated.  

The right information isn’t much use if you don’t have any good actions to choose from. Which brings us to Ashby’s law; “Only variety absorbs variety”. A ‘controller’, (for our purposes a startup), facing a complex situation, must have a range of responses available which matches the complexity of the situation. Otherwise it won't be effective. An experienced chess player has more ‘variety’ than a novice. He is going to employ moves and combinations of moves that the novice doesn’t know how to respond to. Regarding a startup, there is a lot of variety in what customers need and want. To effectively serve the entire market the business would need to deploy a range of products, services and channels. One size fits all won’t cut it. 

For any business, least of all a startup, ‘variety’ in the market is much larger than internal variety. What matters is the ability to match incoming variety with what you can cope with. We talk so much about ICP, product-market fit, and segmentation because it’s Ashby’s law at work. They are all concepts that deal with variety matching. Focus is how you cope with the variety in the market - concentrating on a small area where your capabilities match what is demanded of you. Giving you time to build capability to take on more.

Past, present, and future, all at once

It’s easy to fall into the trap of only thinking in terms of statics i.e. point in time assessments of topics like product-market fit. But growth is a dynamical process through time. Signs of intelligence through time would be to…

1. …control the present, to achieve a certain future outcome

2. …predict the future, given your actions, and those of others

3. …explain the past to better control the present, and predict the future

To explain what has happened, control what you are going to do, and predict the future, you need an internal representation of the world. You need a model. “Every good regulator of a system must be a model of that system” - the Conant-Ashby theorem. Intelligent organisations are therefore adept at building models that drive decision making. As an example, the revenue bow-tie by Winning by Design has become ubiquitous as it is an excellent model for B2B subscription businesses. Ditto the growth in RevOps which uses modeling extensively to build a system of control for go-to-market activities.

Common models/representations found in SaaS and Cloud businesses:

• An ideal customer profile represents your best customers, providing clear guidance over who to target

• Segmentation builds on one narrow ICP by breaking the market down into multiple segments, representing differences in needs

• Customer lifecycle stages is a combined reflection over how customers buy, and how you will sell

• Outbound sequences is a model of the series of steps most likely to lead to engagement and a first meeting

• Sales forecasting leverages historical trends, conversion rates, capacity, and customer lifecycle stages to predict future bookings

Intelligent businesses deal with the past, present and future all at once. Looking back to explain why performance was what it was. Controlling the business to generate target performance now. And predicting what will happen. They aren’t separate phases but parallel processes. They must operate in parallel because everything is changing all the time. The faster a startup can adapt its models based on feedback, the better its control. Information flows combined with speed of adaptation become a catalyst for intelligence.

I have a need for speed

This is what John Boyd noticed in a completely different setting to business. John was a United States Air Force Colonel, fighter pilot in the Korean War and Commander during the Vietnam War. His key insight was that the ability to rapidly and accurately acquire and act on information is a source of competitive advantage. He gave us the OODA loop, a high level abstraction on how to do this systematically. It’s proven its effectiveness and universality, employed by fighter pilots, emergency response drivers, litigators and more. 

OODA stands for Observe, Orientate, Decide, and Act. We’ve already touched on all the components of the approach: information; modeling; control; prediction; and feedback. At the heart of the approach is a model which drives decision making, and predictions over what is going to happen. Feedback is used to constantly update the model and improve how well it reflects the thing you’re trying to control.

Driving is something most of us can relate to and demonstrates OODA at work. You constantly scan the environment taking in information. Plan ahead based on where you want to go and any hazards you need to negotiate. Replan as things unfold in real time. Predict what other cars, bikes and pedestrians are going to do. Good drivers drive systematically. Emergency drivers learn the system of car control which is an OODA loop, enabling them to drive faster and safer.

Similarly, intelligent businesses operate systematically. Particularly in how they conduct customer discovery, build new features, and acquire and grow customers. Systematisation isn’t bureaucratic or rigid. Done right, it speeds up decision making and adaptation rather than slowing it down.

Strategy as the key differentiator

Finally, the picture wouldn’t be complete without addressing strategy. Sound strategy provides direction and intentionality to the models you build and the decisions you make. Humans are good at modeling, calculating, and drawing logical conclusions from a body of evidence. But our species is strategically inept. We slowly poison and destroy the environments we live in, we proliferate weapons that could lead to our extinction, and are brought to a standstill by a virus with a miniscule fraction of our complexity. If strategic incompetence is the norm, it becomes the greatest differentiator between intelligent and stupid businesses.

There are no playbooks for good strategy, but follow great strategists like Hamilton Helmer, Michael Porter, Richard Rumelt, Hiroyuki Itami, and common principles quickly emerge. Power. Fit. Information. Speed. Intelligent businesses make decisions that generate power over time. They stay focused and maintain fit by matching internal capabilities to external demands. They over-extend just enough to keep learning. And they are able to change at least as fast as the external environment.

I’ll end with how David Krakaeur, President of the Santa Fe Institute frames intelligence. Simply, it’s making a hard problem into an easy one. It has three dimensions to it which effectively captures the topics I’ve covered above:

1. Representation i.e. a model/encoding of the environment

2. Inference i.e. the ability to operate/compute.

3. Strategy i.e. using representations and inference to drive action with respect to some objective.

I enjoy writing, but it is a tremendous investment. If you enjoyed this article, please consider sharing it with your colleagues and friends so that it may grow. Thank you.

Share

This month’s article looks at a foundational concept called explore-exploit. A dilemma all organisations, individuals, and living things face, critical to survival and growth.

Subscribe now

Why this is important

Start-ups are learning machines, accumulating information about their environments over time. Putting that information to work is what drives growth, but they face a trade-off between how much effort they allocate to learning vs. making use of those learnings. This is known as the explore-exploit dilemma, relevant to every organisation, individual and living thing. It needs constant management and adjustment over time. Getting it wrong wastes time and resources, and at worst poses an existential threat.

Give me the highlights

• Balancing exploration and exploitation ebbs and flows over time, depending on how much you’ve learned, and how quickly the environment is changing.

• Exploration is disorderly, it isn’t efficient, but action creates useful information that we can later exploit.

• Explore-exploit drives the dynamics of growth; misdiagnosing a lack of growth and then exploring too much or too little can be catastrophic.

A balance that shifts over time

Balancing exploration and exploitation over time follows a predictable pattern1. In the embryonic stages of a new business, product or market opportunity, you know very little. The focus is therefore entirely on exploration. Establishing a basic level of understanding which can then be exploited.

Early exploitation is inefficient but necessary to fuel further learning. Consider a new product launch, and winning initial customers. More effort than you’d like is spent on making things work; that customers are happy. Once unanticipated issues are resolved and crises averted, more effort shifts back towards customer discovery as you become more efficient at exploiting what you learn.

There’s always more to discover; the learning never stops. This is the continuous customer discovery needed even for mature products. The environment also changes over time, which destroys information. So some effort is always reserved to maintain information.

All effort should only ever be allocated to exploitation if a product or business is coming to the end of its life. There is no point wasting effort learning more if you are about to go bankrupt.

The phases of exploration and exploitation aren’t strictly linear; you may need to go back to an earlier phase if you see significant environmental change such as the emergence of disruptive technologies or market liberalisation. It’s also not the case that a business is operating along only one of these curves. It naturally varies by function, product and market. The dilemma is not only relevant to management, but to all individuals and their roles. Not least for a founder who must continue to adapt as the internal environment he is a steward of rapidly changes.

Action produces information

“Action produces information. Just keep doing stuff.”

Brian Armstrong

What a wonderful quote. It gets to the heart of the disorderly nature of exploration. You don’t know what you don’t know. So the only way to uncover the insight you need to fuel growth is to continually bump up against the reality of the world. 

“Unless you have a tolerance for failure, you will never experiment, If you don’t experiment you won’t innovate, and if you don’t innovate you won’t succeed”

Jensen Huang

Exploration requires a high tolerance for ‘failure’. But failure should be cheap. Testing a new product opportunity leveraging Customer Success is a cheap way to test and fail. Investing development time without contact with reality first, and then failing, is expensive.

The larger implication for founders is that burn should be as low as possible during times of intense exploration, this could be early in the life of the business, or when considering a major pivot of go-to-market strategy or proposition.

One of the challenges for founders and operators is knowing when the time is right to shift from disorderly exploration to focused and intentional exploitation. For example, when should they shift from exploring multiple segments to committing to one. This isn’t easy to answer, I think there is an art to it, based on feelings of conviction not just hard data. 

Part of Stephen Millard’s pre-mortems with all new Notion Capital investments is to identify the things the business needs to learn and prove. Reflecting on those two questions regularly is a useful primer to guide explore-exploit decisions. Being able to answer fundamental questions such as those listed below should help to rule out committing to bigger bets when you are not ready.

• Do we know who our best customers are?

• Do we know why our customers choose us?

• Can we identify which prospects will be good customers?

• Can we predict our ability to win a prospect?

• If I increase quota carrying headcount, can I predict the impact on growth?

• Do I know how to onboard customers systematically?

• Do we know what success for our customers looks like?

• Etc.

Up and over the sigmoid curve

We spend a lot of time thinking about s-curves and inflection points at Notion Capital. Growth isn’t linear: it accelerates and decelerates over time. It’s frustrating and surprising in equal measures. A major driver of those dynamics is the nature of explore-exploit. The low growth that precedes explosive growth are periods of concentrated exploration. The inflection points are driven by shifts from exploration to exploitation. Deceleration can signal the need to return to exploration to fuel the next phase of growth.

Danger lies in misdiagnosing why growth is stagnant or decelerating. It could be the result of insufficient exploration or exploitation. Exploit too early and you will burn through runway you can’t afford to waste. Explore for too long and you won’t show the traction you need to. Get it just right, and you’re off to the races.

Intelligent organisations build growth off multiple s-curves, mapping to parallel explore-exploit processes. They manage a portfolio of bets, matching resources to conviction. Through a go-to-market strategy lens this could be pursuing multiple customer segments simultaneously, but taking a different explore-exploit posture in each. Exploiting segments they have high conviction in, whilst exploring lower conviction but high opportunity areas. The strategy drives concrete resource allocation decisions, targets, and differentiated actions. Product, Marketing and Sales all play an important role here, exploring and driving learnings into the business, but it’s the founder’s responsibility to ensure those learnings result in a coherent strategy, striking the right balance between exploration and exploitation.

I enjoy writing, but it is a tremendous investment. If you enjoyed this article, please consider sharing it with your colleagues and friends so that it may grow.

Share

This month’s article looks at the difficult task of monetising data, the challenges data businesses face, and how to move address them.

Dentists with terrible teeth

Monetising data is hard. Hard to package, hard to price, and hard to operationalise. You don’t have good visibility over what customers are using your data for. Once you provide data to a customer via an API or data feed you lose control. God knows where it is and what it’s being used for. It’s a high entropy product. Like milk mixing in your morning coffee. 

Same same but different. Same data, different use cases, different value. Without setting prices that reflect what customers are actually doing with it, it’s very hard to be compensated fairly. This drives another fun challenge for data businesses - customer prices way above and below the official price list.

The Sales cowboys have arrived. And I have sympathy for their pricing creativity. Every customer situation feels like a unique snowflake. Sometimes list prices are far too high for what the customer is trying to achieve, other times too low. So Sales do their thing. Pricing to win deals, and make money. But it’s hard to work out if they are doing a good job, getting the most out of each customer, or just making it up as they go. This ‘creativity’ isn’t limited to pricing, but extends to the product itself.

Data products are malleable in a way that SaaS products aren’t. Customers often only need a specific portion of a data product. Specific fields, granularity, or time periods. But a supplier wants standardised offerings that are easy to operationalise and manage. So there is a miss-match between customer and product that needs to be resolved. As with pricing, Sales can’t help themselves but to ski off-piste, creating Frankenstein data products to meet a specific customer’s needs. This causes havoc for quote-to-cash processes: data businesses are often unable to answer even the most basic questions about their customer base. They don’t know what they’ve sold, entitled, or charged. Sounds unbelievable, but they have the worst internal data about their customers of any business type I’ve worked with. They are dentists with terrible teeth.

A better way

We can’t magic these problems away, but we can move towards a much better world. Firstly, we need to focus on customer use cases, working backwards from the customer rather than forwards from the product. Clarity over what customers use your data for, how they use it, and what value they generate puts us in a much better position to package, license, and price appropriately.

At the heart of the solution lies our ability to build fences between customer use cases, allowing customers to self-select the solution they need, and compensate us appropriately. We can leverage multiple approaches to build effective fences; the most important ones are listed below.

1. How we provide access e.g. GUI

2. What data (fields) we entitle e.g. base data or derived variables

3. What technical features or capabilities we entitle e.g. concurrency

4. What a customer is permitted to use the data for e.g. display

5. Who is permitted to use the data e.g. named users

The most obvious and powerful fence is how access is provided. There’s much less one can achieve with a GUI compared to direct access through an API. However, things fall over if data extracts are permitted out of the GUI, eating into use cases enabled by APIs. A degree of data egress may be inevitable, but allowing too much will be problematic.

Certain use cases may only be possible with specific data fields, or other technical capabilities such as API concurrency. If this is the case, we can configure different flavours of APIs, enabling different use cases, and differentiating prices accordingly. The beauty of this is that customers will self-select which API they need and pay accordingly. It won’t be possible for them to use a low value API for a high value use case as important data or technical entitlements will be missing.

The above examples rely on building physical fences through data entitlements, the access medium, technical features, and so on. It’s where we should start, but unfortunately physical differentiators won't always be available or viable. To be truly effective we need to move beyond the physical to contractual. Successful data businesses such as the LSE, NASDAQ, and LME, leverage creative licensing and the threat of audit to differentiate prices and keep customers honest. If you study these businesses amongst others you will quickly discover common approaches to their licensing, and in many cases common language. The most common license types are listed below.

1. Display license: Permits distribution to colleagues via an internal system

2. Derived data license: Permits data to be used in a formula to create derivative data

3. Distribution license: Permits distribution of data outside of the organisation

4. Non-display license: Permits the data to be used for algorithmic trading

5. Reference license: Permits data to be used as a reference in a contract

In addition to differentiating license types by use case, you will also need to license who can consume your data. Given that you cannot track who is using your data within a customer, contractual mechanisms, although by no means foolproof, are the only practical way forward. For display use cases this would require a list of named direct or indirect consumers of the data, which would then drive the total contract value.

Regardless of the fencing approaches you adopt, you will need to design your product menu thoughtfully, balancing customer requirements and your objectives. An important concept when packaging data assets is the ‘minimum sellable unit’. This is the smallest unit of data that you are willing to sell. Customers are going to push you towards a more granular minimum sellable unit so they can cherry pick exactly what they want. But this drives complexity and small deal sizes. It’s hard to give clear direction on how granular you need to be as it varies by business. Construct a menu you believe is reasonable, and then adjust based on customer feedback and your objectives. However, wherever you land, Sales have to stick to the menu and not deviate without good reason. Otherwise you’ll quickly lose control over your product, pricing, and quote-to-cash processes.

By starting with customer use cases and working backwards, building physical fences, leveraging licensing creatively, and then setting differentiated prices we can alleviate many of the challenges I highlighted at the top of this article. There will inevitably still be challenges you will continue to grapple with, but I promise you’ll be in a much more controlled and profitable place, with happier customers.

Share

This month’s article in my foundations series digs into fit. Why it’s much more than just product-market fit, how it relates to information, and why it’s easy to go down the wrong route and get stuck.

Why this is important

We obsess over product-market fit. But fit applies to the whole business, reflecting its likelihood to survive and thrive. Only a few combinations of resources, capabilities and focus will produce any growth, with even fewer producing exceptional outcomes. Unfortunately it’s not always clear which decisions will lead to an increase in fit; the environment can also change and destroy it.

Give me the highlights

• Fit for purpose: Fit depends on the relationship between a business and it’s environment; not only on the product.

• Information drives fit: The more you know about your environment, the better predictions you can make, and the better your fit.

• Designing for the unknown: The more you explore, the better decisions you can make, but this has to be balanced with exploiting what you know.

• Climbing the wrong hill: Most combinations of capabilities and focus are false summits, requiring a step back, or a creative jump to drive better outcomes.

• Dealing with earthquakes: Environmental changes destroys fit - the path out requires a return to exploration and experimentation.

  ---

Fit for purpose

Survival of the fittest has nothing to do with fitness and everything to do with fit. Try driving an F1 car in a desert and you won’t get very far; take it to Silverstone, and it will show you what it can do. Performance depends on the relationship between driver, car, and track. The same goes for any business. Its performance is a function of how well it has been designed for its local, immediate environment.

Product-market fit has existed in the venture capital lexicon for decades, go-to-market fit, and scale-to-market fit entered more recently. The industry keeps having to add more variations of ‘fit’ because fit is an emergent property of the whole; something different from the sum of the parts. As with any design, improving one part is likely to force trade-offs elsewhere. Increase the functionality of your product and you may need to also increase the level of support.

Each location on the landscape above corresponds to a different configuration of the business in a given environment. For example, different technologies, focus, human capital etc. Height represents reward. The ups and downs of the landscape are a function of complexity and interdependencies within the organisation, and with its environment. Moving through the landscape, fit might initially decrease, but new invisible assets accumulate which helps it to climb higher peaks. Build a business on the edges and fit is elusive. Move towards the centre and things improve. The green path up is smooth; take the red path up and you’ll make it to the top with some bumps along the way.

Growth is a step function because small changes can lead to a significant improvement in fit. In the landscape above, a peak is created by a unique combination of customer and product focus. You may operate very close to that peak, where it is rather flat, but a small change in customer and product focus unlocks a sudden jump in fit. I see these jumps happen often. Small changes in the ICP, or small additions to the feature set can lead to significant improvements in fit.

“The value (“fitness”) of a given combination of building blocks often cannot be predicted by a summing up the component blocks.”

David Krakaeur

Information drives fit

Maxwell’s demon can bring order to chaos. The demon appears in a thought experiment where he controls a trap door between two chambers of gas. He survey’s the environment, measuring the speed of the molecules, and then uses that information to separate slow moving molecules from fast. His ability to build up information about the environment is what gives him the power to create order. The demon isn’t just a thought experiment; we can find him in the natural world. He encodes information about the environment into DNA through random variation and natural selection. The demon chooses what to keep and what to discard, improving genetic fit over time.

Information, which I wrote about here, and fit are analogous. Information about the market is encoded into the business; lots of small refinements in its processes and product drives fit. It’s not about market reports and strategy docs, but how the business interacts with its external environment. Businesses that know a lot about their environment can also make predictions better than chance. They know how likely they are to close a deal, how to make their customers successful, what features to build, and so on.

“Information and fitness are exchangeable. You have high information about how to do something, how to survive in the environment means having high fitness.”

Christoph Adami

Designing for the unknown

Designing a start-up to fit its environment is a bit like an F1 team building a car when it doesn’t know the driver, the surface of the track, how long it is, or the number of laps. Considerable effort needs to be spent on understanding the environment before making consequential design decisions.

Founders and their executive teams face a dilemma. Exploring the environment allows them to learn more and make higher quality decisions. But explore for too long, and they won't show the traction needed to secure additional capital. Exploit too early, and they could end up in a dead-end or burn through runway with nothing to show for it. Jeff Bezos framework of one vs. two-way door decisions focuses exploration effort where it matters. Some decisions are so consequential and costly to reverse that they deserve to be over-analysed. But most decisions are two-way door decisions, allowing more freedom to navigate the landscape and find the best fit experimentally on the understanding the decisions are reversible. Progressively building up a map of the territory is another way of limiting how much you need to explore whilst still making high quality decisions.

You don’t need to know much about the London underground to navigate it. Just a tube map. A simplified abstraction of the territory. Just as a tube map is helpful to decide what route to take, market segmentation guides decision making and strategy. It saves you needing to explore every potential customer in the market, instead capturing the differences that really matter. Focusing on one segment of the market (your ICP), provides the additional advantage of reducing the size of the landscape. With less to explore, you’re able to focus your efforts, and learn faster. The typical sequencing therefore proceeds as follows: broad exploration —> build a map of the territory —> focus in on a narrow ICP —> progressively increase exploitation —> explore the next opportunity and so on.

“Some decisions are so consequential, and so hard to reverse, they are one-way door decisions. You go through that door, you’re not coming back.”

Jeff Bezos

Climbing the wrong hill

A group of Navy SEALS were discovered by local Taliban sympathisers in the Korangal Valley. Instead of killing them, they let them go, knowing that this would jeopardise their mission. Unable to update their superiors, they hoped that by climbing a peak they would be able to call for help. Their hearts sank on reaching what had looked like a peak, only to discover a false summit where their communication equipment failed to work. Unable to escape or call for help they were overwhelmed by Taliban forces. Of the four SEALs, only one survived. The difference between climbing a local vs. global peak was the difference between living and dying. This is the true story of Marcus Luttrell, as depicted in the book and film adaptation ‘Lone Survivor’.

The hill a start-up chooses to climb might not have life or death consequences, but it is the difference between a great and a mediocre outcome. Given that you don’t know what the landscape looks like, progress can be misleading. You keep climbing higher, only to get stuck at the top of a small hill where inertia is at a maximum. Whichever way you step is a step down. There are few options available at the top: i) stay where you are and accept a mediocre outcome; ii) take the pain of going down and find your way to a higher peak; iii) if you’re lucky, take a creative leap across the landscape onto a higher peak.

A creative leap is likely to be driven by a change in ICP and application of your invention. Sildenafil i.e. Viagra was originally intended to treat hypertension. After noticing side effects I don’t need to elaborate, Pfizer targeted a new condition where fit was unexpectedly excellent, creating a multi-billion dollar per year product - a creative leap across a landscape.

"Bad! Bad! What? Is he not going - back? - Yes! But you understand him badly if you complain about it. He's going back, as every man does who wants to make a huge jump.”

Friedrich Nietzsche

Dealing with earthquakes

‘Only the Paranoid Survive’ is a book about earthquakes. The kind of earthquakes that upend entire industries. Significant environmental change destroys hard won fit. It’s as if you climbed Mount Everest, only for the mountain to collapse, and then a larger mountain to emerge in the distance. When the environment changes there is no choice but to return to exploration, to loosen control. New invisible assets need to accumulate to return the business to a position of fit.

Loosening control provides colleagues deeper in the organisation space to experiment and learn - building up the information needed to chart a new path forward. Andy Grove successfully led Intel through multiple earthquakes by ‘letting chaos reign’ at the right time. This is why retaining a  culture of experimentation is so important. As the environment changes experimentation needs to be dialled up, but that’s hard to do if it isn’t already a feature of the business.

An underlying culture of experimentation provides the seeds for future adaptation. AWS grew out of a desire to increase the internal efficiency of Amazon. Intel already had a small microprocessor business before doubling down. In both cases early exploration and accumulation of visible and invisible assets set the business up for success as their environments shifted. Hard questions also need to be asked about how the business is configured. The most sensitive will concern people. Does the team have the right knowhow to thrive in the new reality? Can they acquire the required knowledge and skills, or do people with the right knowhow need to be brought in?

“It is not the strongest of the species that survives, not the most intelligent that survives. It is the one that is the most adaptable to change.”

Charles Darwin

Why this is important

Startups are learning machines. They come from a place of low information, not knowing who their best customers will be or how to make them successful. To grow, a startup first needs to accumulate information. Not in PowerPoint slides and excel sheets, but physically in its product, brand, and operations.

Give me the highlights

• Information is physical: Information isn’t just what you know, it’s encoded into your business, and takes time to accumulate

• Information flows: Information flows from the market, to the market and internally - the capacity of information to flow and accumulate are key enablers of strategy

• Information emerges from disorder: Information emerges at the edge of chaos, making disorder an important feature for adaptation and growth, not an error to be removed

• Information creates order: Information creates order and order is needed to grow - judging when to dial disorder up or down is a key role for founders and CEOs

• Information accumulates before revenue: A startup’s search for greater fit between its product, sales motions and the market, and the accumulation of information are two sides of the same coin - they happen before growth ignites

Subscribe now

Information is physical

If you’ve visited the beautiful city of Prague, home to Notion Capital portfolio company Mews, you may have seen the base of the statue of St John Nepomuk depicting a knight petting his dog. Information is encoded in the scene. There is the story of the knight and his dog, and then there is the story of how passers by have interacted with the statue. Over many years of petting the knight’s loyal friend he shines brightly compared to the rest of the bronze scene. That is information. The physical objects we create, each with a history of interactions, not just what we know.

Only finite information can be stored in you and your colleagues' heads. For information to grow, which is what you need to happen if you want revenue to grow, it has to be stored physically somewhere in the business. Your product encodes information gleaned from your customers, and your sales processes and enablement material encodes what you’ve learned about how to acquire your ideal customers. The greater the depth of information a business is able to accumulate, the better placed it is to adapt and thrive.

Information evolves from simple to complex, it doesn’t jump from zero to one. It took 4.5 billion years for humans to evolve into what we are today. We didn't just spontaneously appear. The same goes for encoding what the business learns physically. Processes start simple, and evolve gradually over time based on feedback loops. Sophisticated products and processes aren’t just rich in information, they are deep in time. Knowing the next step of evolution from someone that has been there and done it is helpful. On the other hand, trying to replicate mature structures in a young business is a recipe for disaster. Respect the process, there are no shortcuts. Evolution may be slow, but it's relentless.

“Physical order, or information, is what is embodied in a product.”

Cesar Hidalgo

Information flows

Information flows from the market into your business (environmental), from your business into the environment (corporate), and internally (operational). As we’ve seen above, information has to be saved somewhere to accumulate. The capacity to bring the right information into the business is a valuable asset, and the choices over which flows to build are strategic.

In the start and build phases of growth, focus should be much more on accumulating information than growing revenue or efficiency. Prioritising ‘efficiency’ over information flows can be really damaging to medium and long-term growth prospects. Primarily because being ‘efficient’ is often a local not global optimum - a topic for another day. Two concrete examples I want to highlight, demonstrating the need to think bigger than short term efficiency: i) outsourcing the BD function and; ii) product-led growth. 

Outsourcing an BD function might sound like a great idea from an efficiency perspective. It can be spun up quickly, it's flexible, and you don’t have to go through the pain of hiring. But is an outsourced BD function a high capacity flow of information from the market into the business? No! The ability to accumulate information (remember information is physical) is going to be diminished.

With respect to PLG, it’s all the rage. Sorry to rain on the parade, but is a self-serve motion where you never talk to your customers a high capacity information flow? No. I’m not saying PLG doesn’t make sense, but it has a cost. At the very least, if a new business is operating a PLG motion, it should be reaching out to every early customer to learn more. Something Aikido, a software security platform has embraced to maximise learning whilst operating a PLG motion.

“Capacity to bring information into the firm, handling the flow of information generally is an important invisible asset.”

Hiroyuki Itami

Information emerges from disorder

I come from an ordered world. A strategy consultancy, German no less, full of two-by-two frameworks and neat models. It’s Notion Capital that has exposed me to the value of disorder - a useful feature of a startup not a fault. The disorder I am referring to is not random. It’s chaotic but there is a logic to it. Logic that becomes increasingly ordered as information slowly accumulates and strategy is refined.

Think about your search for product-market fit. You want to find the customers that your product resonates with now, but also the next segment to target. So you initially try to sell to a broad range of potential customers to learn. You soon find that the customers your proposition resonates with differ from your initial hypothesis. This is often the case. You simply know less about the market than you think you know. Your disordered exploration of customer types is what provides the insight you need. Insight that would have alluded you with an overly ordered approach.

In Andy Grove’s ‘Only the Paranoid Survive’ he talks about the dangers of experienced operators implementing the same strategic and tactical moves that worked historically for them. They are unwilling to stray from their well travelled paths. There is no disorder in their approach. They fail to appreciate that this time, things might be different. His advice is to loosen control, to let ‘chaos reign’, especially as a business tries to navigate an inflection point. But this culture of experimentation needs to be an existing feature of the business which can be dialed up or down. It’s not something that can be kicked off from a standing start. 

“Information is hidden on the other side of chaos”

Ilya Prigogine

Information creates order

As information accumulates bottom-up, founders need to start dialing back disorder and drive greater focus and control top-down. Strategy isn’t top-down or bottom-up, it’s both. A balance that needs to be dynamically adjusted through time. Responsibility for getting this balance right sits squarely with the CEO/founder.

We have a saying at Notion Capital “get comfortable with being uncomfortably narrow”. We have a healthy respect for the disordered nature of startups, but there comes a time when you need to focus and place your bets. That’s the only way you can generate leverage from the limited resources you have. Efficiency becomes increasingly important over time, and that means less exploration and more exploitation of the invisible assets you’ve accumulated. It’s at this point where it makes sense to start deploying capital more aggressively and go for growth.

“We grow in direct proportion to the amount of chaos we can sustain and dissipate”

Ilya Prigogine

Information accumulates before revenue

Every set of circumstances is unique. Founding teams vary in their depth of information of the market they wish to serve. Some come equipped with detailed maps, others with a vision and general sense of direction. Regardless of the starting point, information accumulates before revenue.

Information and fit are equivalent concepts, and both are relative to the environment. Solutions that are deep in information are a better fit for their target problem. It’s the movement towards fit in multiple domains that drives growth over the medium term. And it's these periods of low fit when information accumulates the quickest. The more complex the market, the more information in absolute terms needed to achieve fit.

Because information is relative and context dependent, it can be destroyed when the environment changes. Imagine an up-to-date map of your local area. It’s informationally rich. However, that same map in fifty years is likely to be much less useful. As the external environment changes, parts of the map become irrelevant, it’s as if some information encoded in the map has been destroyed. The gap between potential information (a perfect representation of the territory) and actual information (the map) increased. Or said another way, fit declined.

Consider entry into a new market. You may think that your proposition suits a new geography but find that fit is much worse than initially assumed. Growth eludes you. But if you have the right flows set up, and can accumulate and encode information from the market into your proposition, you will improve fit and eventually ignite growth. It’s interesting to reflect on the success of US market entry when a founder moves to the US vs. hiring a local sales leader. Anecdotally the former is much more successful than the latter. One reason for this could be that a sales leader is there to sell, whereas a founder is much better positioned to drive information into the business and improve fit.

Growth isn’t linear. It zigzags precisely because a business thrusts itself into states of low fit. The lack of fit combined with the right flows causes information to accumulate. This takes time. But once a critical threshold in information/fit is reached, growth ignites. 

“It is the accumulation of information and of our ability to process information that defines the arrow of growth”

Cesar Hidalgo

Subscribe now

Highlights

• Platforms act as intermediaries between participants, managing transactions and/or network effects

• Platforms are nothing new, but have become embedded in our daily lives and are impossible to avoid

• They create value by removing friction, coordinating individuals and businesses, and facilitating trade

• One of the key roles of a platform is to manage network effects between participants, which is achieved partly through a well designed pricing strategy

• Not all platforms are destined for profitability even at significant scale, they require economies of scale to generate market power

  Subscribe now

  ---

What is a platform business?

Platform businesses are nothing new. There are examples stretching back thousands of years from ancient China, where meiren (matchmakers) would orchestrate marriages and alliances between families, to trade fairs in the County of Champagne1 where participants were carefully selected, mechanisms put in place to enforce contracts and resolve disputes, and credit provided to facilitate trade, all in return for the Count taking a share of the transaction.

Fast forward to today, I am writing this on Google Docs, through Google Chrome, using a Microsoft Windows laptop, sending and receiving data through Notion’s ISP. That’s at least four platforms I’m leveraging to write a single article; a fraction of the true complexity and layers upon layers of platform businesses we are all dependent on.

In an analysis of the Forbes Global 2000 between 1995 and 20152, platform businesses achieved operating margins of 21% vs. 13% for non platform businesses; sales multiples of 5.4 vs. 1.4; and sales growth of 18% vs. 8%. As of January 20243, seven of the top fifteen most valuable public companies in the world based on market capitalisation operate significant platform businesses. For some, such as Microsoft, Meta, VISA and others, platform thinking has been fundamental to how they create value rather than complimentary to a traditional business model.

Despite the success stories, platforms are not guaranteed to become profitable businesses that endure. They have more complex dynamics than traditional businesses and employ what can seem like counterintuitive strategies especially with regards to pricing. The French economist Jean Tirole won a Nobel Prize4 in 2024 partly due to his discoveries into multisided platforms. But before we dig deeper, it's useful to define what exactly we mean by ‘platforms’.

In ‘The Economics of Platforms’ Belleflamme and Peitz define platforms as an “entity that actively manages network effects between economic agents”. Cusumano et al in their book ‘The Business of Platforms’ also require the presence of network effects, splitting platforms into either facilitating transactions and/or acting as a technology foundation for the development of third party innovations.

Given that network effects can be weak, negative, and hard to identify, I prefer a broader definition that does not depend solely on network effects but also includes businesses which act as intermediaries for transactions regardless of the presence or materiality of any network effects. 

“A platform acts as an intermediary between participants, managing transactions and/or network effects.”

How do platforms create value?

Platforms aren’t ‘normal’ businesses. They don’t necessarily have a product to sell and don’t always own the relationships with their participants. So how do they create value?

They primarily solve for market failures and frictions where a degree of coordination between people or businesses is required. They facilitate cooperation where it would otherwise fail, reduce search and transaction costs, and provide safe spaces for participants to transact in confidence. I’ve provided some more colour and examples below; the list of benefits is not necessarily exhaustive, but covers the main vectors of value.

Coordination: “I am able to cooperate with others in ways I fundamentally wouldn’t be able to do on my own”

The Interbank Card Association and VISA networks were set up specifically to help acquiring and merchant banks communicate and settle card transactions. Without an intermediary it wouldn’t have been possible for a bank to develop all the relationships it needed to.

Matchmaking and search costs: “I can find what I am looking for”

LinkedIn helps people to connect with that exact individual they’ve been looking for, whilst Uber helps riders to find a car at the right time, and OpenTable to find an available booking for a Friday night. All of these platforms facilitate matchmaking by reducing the cost of search and improving the quality of matches.

Communication: “I can communicate with the participants I’ve found”

Finding what you’re looking for is often just the start of the journey. Platforms such as WhatsApp's fundamental proposition is to facilitate communication, whilst Ebay and Etsy provide the ability to communicate so that buyers and sellers can gain the clarity and confidence they need to transact.

Transaction: “I can complete a transaction efficiently”

Once you’ve found what you’re looking to buy on Etsy, you’ve asked clarifying questions, and gained a level of confidence in the seller, you then need a way to transact efficiently and safely. Platforms are well placed to sit directly in the payment flow, earning the right to a share of transactions. Being able to transact directly on a platform is great for the platform itself but also users, providing a seamless user experience and additional assurances.

Trust: “I can trust this person, or the platform can provide me assurances”

Providing trust plays an important role across many platforms and throughout the lifecycle of engagement between participants from search to eventual transaction. Platforms can provide a level of trust directly through assurances such as eBay’s money back guarantee and Alibaba’s escrow functionality, or deploy features such as identity verification and ratings to weed out bad actors. More generally they determine the governance rules that moderate participant behaviour, acting as a trusted neutral party to resolve disputes.

Managing network effects

A lot of what I’ve covered above depends on network effects, which reflects the impact of adding an additional participant, or an increase in activity on the value of the platform. Network effects will vary in strength and consequence for different groups. A within-group effect reflects the impact of an increase in participants or activity within the same category of participants e.g. members on LinkedIn, whilst a cross-group effect reflects the impact of an increase in participants or activity on a different category of participants e.g. an increase in the number of restaurants available on the OpenTable platform for restaurant goers.

Network effects are not exclusively positive, they are often negative. Traditional businesses sell ‘goods’, platforms often bundle in ‘bads’ intentionally. Advertising is a common ‘bad’ leveraged by platforms. Users create a positive network effect for advertisers by providing their attention, whilst advertisers create a negative network effect for users by supplying advertising they would rather do without. At other times the platform needs to mitigate the impact of damaging network effects it didn’t want. For example, when you search for a ride with Uber, you would rather not have to compete with other riders. They are imposing a negative within-group effect on you. Uber needs to make sure it manages supply and demand appropriately so you don’t get frustrated and switch to a competitor.

One of the key functions of a platform is to manage network effects whether they are positive or negative, intentional or unintentional. How they design their platform, the features they build and governance policies they implement all play an important role. But one of their most powerful levers is pricing. For a traditional business, the prices charged have no impact on the usefulness of the product it supplies. For many platform businesses the prices charged or as we will see subsidies provided have a direct and profound impact on the value and viability of the platform.

Pricing for platforms

Getting the pricing strategy right for a platform is much more important and complex compared to a traditional business. I’m simplifying, but a traditional business really just needs to understand how sensitive customers are to changes in price, whilst a platform needs to worry about how changes in price will impact network effects, and the overall value of the platform.

Key questions for a platform as it designs its pricing strategy:

1. How much value does the platform create in aggregate across all groups?

2. What’s the minimum value each group requires to participate?

3. How sensitive is each group’s participation/activity to changes in price?

The first question deals with whether the platform is addressing enough friction to be viable. The greater the value pie, the more there is to share amongst participating groups. Second, it’s possible that one side of the platform might need an additional incentive to participate, whether that’s being provided something valuable for free or direct payments. Third, each group's participation will be impacted by changes in prices/subsidies, which in turn impacts the value of the platform in aggregate.

Pricing decisions can therefore dynamically alter the value of a platform by impacting participation or activity levels of one or more sides. Getting the balance right will also change over time as the platform proposition develops, or the competitive landscape and negotiating position of participating groups changes. I’ve provided a summary of pricing strategies below for some well known platforms to show how network effects have impacted their approach.

American Express: American Express pays cardholders to use their card either through cashback or rewards. These payments are funded by merchants who pay various fees to AMEX for bringing them customers that tend to spend more than competing card networks, as well as cardholders who roll their credit into a high interest loan. Given that individuals choose which card to use, it makes sense for AMEX to incentivise usage, paid for by the merchant side.

PlayStation/Xbox: New generations of games consoles typically sell below unit cost as they try to build scale (demand side). Sony and Microsoft also provide significant tooling and support to developers for free to help stimulate the development of games (supply side), which is all paid for through royalties generated from game purchases. Getting the price wrong for the console can be really damaging to a new generation's prospects as Sony discovered by overpricing the PS3, and Microsoft found when it hard bundled an expensive camera with the Xbox One.

LinkedIn: LinkedIn’s value is in its scale. The ability to connect with anyone across a broad range of industries has proven to be extremely useful. It offers paid versions with more functionality but that isn’t where it makes its money. The majority is generated through its Talent and Marketing Solutions, both of which require the attention of professionals at scale to recruit from and market to.

The power to grow profitably and endure

Businesses with market power are more fun to work for than those without. It’s what enables them to grow market share and margins vs. competitors and endure through time. The economist, advisor and investor Hamilton Helmer has written more clearly about the properties of market power than anyone else in his book 7 Powers. How power relates to platforms is something he is actively exploring, with some of his early thoughts shared in the Acquired Podcast - well worth a listen.

I’ve summarised his framework below and simplified some of the language, as we can make use of it as we explore what conditions need to hold true for a platform to have a shot at generating power.

Two of the seven powers jump out as being particularly relevant to platform businesses: i) network economies; ii) scale economies.

Network Economies

Hamilton identifies network economies explicitly as a source of power, inherent to many platform businesses. A naive reading of this would be that businesses with network effects enjoy market power, but this isn’t the case. Easy to miss, but it’s reflected in his use of ‘economies’ rather than ‘effects’. Network effects need to be material and difficult for competitors to replicate to generate power. One of the most likely barriers is whether network effects grow in value as the platform scales.

Scale Economies

As the scale of a network increases, it’s likely that the quality and efficiency of matching increases and hence so does the value of the network. There may come a point, however, when increasing the size doesn’t have any impact on either value or cost, or it actually starts to have a negative effect. Think of a nightclub you went to too early in the night. It starts off empty and is a pretty boring experience. As it fills up things get more fun for everyone up to a point. Beyond that, it becomes an awful experience as it takes ages to get a drink at the bar and there’s no space for you to show off your terrible moves on the dance floor.

Every platform is different, some are lucky and stay on the ‘increasing’ part of the curve forever, others start to experience constant returns faster than they would like, and if you are really unlucky, they’ve got to deal with damaging decreasing returns or limit the size.

Profiling power

When I set out to write this article I wanted to end with a simple framework that could help categorise platform businesses based on their ability to generate power. That’s proven to be much harder than I imagined. After many iterations I’ve landed on the below which focuses on three things: i) scale economies; ii) nature of network effects; and iii) whether participants are likely to use multiple competing platforms.

Platforms with increasing economies of scale are always going to have more power than those without. Those that benefit from meaningful cross-group network effects are likely to have solved tougher coordination problems than platforms that benefit predominantly from within group effects. Finally, it’s preferable if participants can’t multi-home (participate in multiple networks at the same time).

Just because a platform falls into one of the ‘low power’ buckets doesn’t mean it can't generate market power, rather it’s an indication that it will need to rely on other sources such as switching costs or brand.

I’ve mapped out a few logos below based on where they fit best, however, this does require a degree of judgement. I set a high bar regarding economies of scale; all of these businesses will have enjoyed economies of scale to some extent but what matters is whether that continues over significant scale. I’ve also included the latest operating margins to give some idea of performance.

It’s challenging to find platform’s which benefit from economies of scale and predominantly within-group network effects. It may be that once a platform achieves scale it tends to either get acquired or adds an additional side to the platform to drive monetisation. For example, Facebook’s acquisition of WhatsApp, or LinkedIn’s addition of advertisers and businesses to capitalise on member attention.

Apple and Microsoft are enormous businesses with multiple platform businesses within them. One could make arguments as to why they should move down a bucket, or that different business units should belong in different places. I’ve profiled them as ‘single home’ due to the fact that you only ever really have one operating system installed on your computer, although you are likely to own multiple devices with different operating systems. Whether we push them down a bucket or not, all four logos in that column have significant market power, due in large part to economies of scale.

All the examples I picked on the left enjoy economies of scale and generate impressive margins. The right hand side is a much more mixed picture. Many of these businesses operate at significant scale, but none of them generate meaningful operating margins with ¾ of the examples losing money. A lot is obviously going on under the surface to generate these figures, but they are interesting comparatives nevertheless. An interesting next step is to profile a much larger set of platform businesses to see if any clear patterns emerge.

Summing up

• Platforms act as intermediaries between participants, managing transactions and/or network effects

• They create value by removing friction, helping participants to coordinate and transact safely

• Network effects play an important role, but they are not always positive, and do not necessarily keep improving with scale

• Pricing plays a critical role in managing network effects and ensuring value is shared amongst participants in a way that maintains the value of the platform

• Not all platforms are destined for market power, they require economies of scale at significant scale to generate healthy margins

This has been a far more challenging article to write than I anticipated. Platforms have much more complex dynamics and employ more novel strategies to win than traditional businesses. As software continues to eat the world, it will be interesting to see the business models they disrupt next. B2C has experienced the lion's share of creative destruction so it will be interesting if we see it is more weighted to B2B platforms in future.

References and additional reading:

• Matchmakers by Evans and Schmalensee

• The Economics of Platforms by Belleflamme and Peitz

• The Business of Platforms by Cusumano, Gawer and Yoffie

• 7 Powers by Helmer

Formalising Product-Market Fit

The thinking behind how products create value across customers stretches back to founder and legendary Sequoia Capital investor Don Valentine, which Andy Rachlef of Benchmark Capital christened as product-market fit. 

In our Start, Build, Scale framework, achieving a high degree of product-market fit should be the primary objective of the start phase before moving onto building a repeatable go-to-market engine and investing in growth. 

The term is now so widely embedded in the start-up ecosystem that it’s almost impossible to have a conversation without it coming up. In reality product-market fit is much messier than the way it is typically spoken about, which can unfortunately lead to some very difficult situations for both founders and investors which we will touch on in this article.

My intention here is to formalise the definition so that we establish clarity over what it is and isn’t, and to ultimately drive better decision making. There will be variations in the words people use to define it, but essentially comes down to creating value for customers in a way that is repeatable for a given market.

Five things to know about product-market fit:

1. Product-market fit is a function of customer value created, not value captured

2. Product-market fit is a function of market scale

3. Product-market fit requires repeatable value creation

4. There are only ever degrees of product-market fit

5. Product-market fit is tightly bounded to common customer and market characteristics

   Subscribe now

   ---

Axiom 1: Product-market fit is a function of customer value created, not value captured

Invention is the mother of product-market fit. Peter Thiel argues that inventors should be aiming to create technologies that are 10x better than alternatives. I think his definition is too restrictive, focusing only on technology, when there are other sources of innovation, but his fundamental point of value creation sits at the heart of product-market fit.

Importantly, value creation has nothing to do with the prices that suppliers charge or the margins they earn. Being able to charge high prices is an indication that you have created something valuable for someone, but it is not a requirement for product-market fit. Your ability to capture value depends on your ability to first create value, but the logic doesn’t flow backwards i.e. your ability to create value for customers doesn’t depend on your ability to capture it. That depends much more on industry economics and competitive position.

“Price is what you pay, value is what you get”

Warren Buffet

Take a simple example: two situations for buying a chocolate bar i) a busy street with many shops ii) a train with one shop. Your willingness-to-pay is different because it is shaped by competitive forces, the value you get is the same (assuming you are about as hungry).

Action for founders: Invest time in understanding the magnitude and nature of value your solution drives for customers, including the economic value to the corporate, social value to employees and stakeholders, and psychological and emotional value to the buyer. Finally, don’t confuse price for value, even though it can be a good indication of having built something valuable.

Axiom 2: Product-market fit is a function of market scale

For a business to achieve significant scale, there is no use creating something immensely valuable if it is only valuable to one customer. It’s in the name, but product-market fit depends on creating value for a (large) market of customers. As markets vary in size, product market-fit is a function of market scale, which one can think of simply as the number of potential customers you could sell to based on your market definition.

The shape of product-market fit will be idiosyncratic to the invention and target market. A company selling an operational risk management solution into tier one banks in the UK will only have a handful of potential customers, but will create immense value for those customers. Whereas, a business selling an email service solution to small retailers may have tens of thousands of customers, but will only create modest value for each customer. The total area for these respective businesses may be the same but the shape will be totally different.

Action for founders: Invest time in getting a sense for market scale, and how that can be broken down by different needs-based segments. There is no need to do a large expensive market study early on, but you should have a good feel for the scale of the market and where the value pools are.

Axiom 3: Product-market fit requires repeatable value creation

Axiom 2 infers repeatability but it’s worth calling out separately as it’s so fundamental to the definition. The value that a new invention/proposition creates has to be repeatable across the market i.e. the invention has to fit a market as a whole not just individual customers. Founders are in the business of building repeatable, scalable products where the same product can be sold to multiple customers, not where every customer sale is a bespoke project.

Building highly customised propositions can generate significant revenue in the short term, but isn’t particularly scalable and will act as a drag on growth. Early stage investors, including at Notion Capital sometimes invest in a start-up partly on the basis that it has achieved a high level of fit for a marque logo. The assumption is that success can be replicated in logos which look similar. Unfortunately this is not always the case, and to create repeatability the business needs to either pivot its solution or the market it’s targeting.

Action for founders: Do not assume that just because you have created something valuable for a large customer that is paying a lot, that your solution will create value in a repeatable way for similar looking customers. Get to the bottom of which customer needs are driving fit and don’t rely on misleading descriptive customer characteristics. Test repeatability with similar customers before investing in go-to-market.

Axiom 4: There are only ever degrees of product-market fit

Hopefully you have picked up by now that since both value and market scale are continuous variables, so is product-market fit. We can only ever talk about degrees of product-market fit. Statements like “we’ve achieved product-market fit” are unhelpful and can be misleading and potentially dangerous as it can lead to investments into go-to-market built on shaky assumptions. Shortcomings could be driven by the value side (not creating enough value for customers) or due to market scale, which as we will see in axiom 5 is sensitive to how precisely we define the market. It’s also important to remember that the world is dynamic, not static, so the value created for customers and the scale of the market can and will shift over time. Product-market fit is a shifting phenomenon: it shouldn’t be thought of as fixed for the life of the business.

Action for founders: Stamp out language that implies product-market fit is a binary outcome. Be very precise in your definition of segments and markets, and do not treat product-market fit as something that is fixed through time.

Axiom 5: Product-market fit is tightly bound to common customer and market characteristics

Market definitions are really tough to get right, but the degree to which you achieve product-market fit depends heavily upon it. Defining the market too broadly can give a false sense of scale, implying fit where it does not exist. Product-market fit is tightly bound to markets/segments, believing you have a high degree of fit in one market/segment does not mean you will achieve the same degree of fit in another market, which one could think of as different customer verticals, geographies etc. Sounds obvious, and it is, but it still catches investors and founders out.

Think of it as absorbing barriers between markets (a point you reach beyond which you cannot continue), you cannot necessarily extend fit from one market to the next without changing the product. For example, in the diagram below, investing in go-to-market motions in market one based on the success you have in market two could be an extremely costly mistake, a mistake we’ve seen many times.

Action for founders: Segment the market based on true differences in customer needs, which reflect different degrees of product-market fit. Evaluate product-market fit in different segments to help determine go-to-market sequencing and ensure you are not over-investing in growth where you lack product-market fit.

Progress: Gradually then suddenly

Imagine a block of ice in a cold room. Your objective is to melt the ice, but you have imperfect information about the temperature of the room. You are able to warm the room up, but it’s costly, so you progress gradually not to overshoot. You increase the temperature by 1 degree at a time and observe what’s happening. Nothing, no reaction. You’re confident you’ve progressed towards the melting point, but you still don’t know how much further there is to go. You pass that magic zero degree point. Still nothing happens. The room is now warm enough but you need time to see a change. You’ve passed a threshold in this simple system, but have no idea change is happening. So you keep increasing the temperature, overshooting by a few degrees above zero. Finally you start to see something happen. Some evidence of melting. This is the inflection point. And if you wait long enough the ice will melt completely. If you were to plot this process on a graph from -5 degrees to 5 degrees you would see a step function, from frozen to liquid. On a graph it would look like a sudden and dramatic change, but in reality your experience of it would have been a mixture of uncertainty, frustration, gradual progress and then sudden change. This is the essence of a startup’s growth journey.

As a startup grows the nature of the challenge changes, and so it must change if it wants to continue growing. Founders will see inflection points in growth indicating both the need to change, and the result of change. In-between are the critical thresholds that must be overcome, and the reinvention needed to unlock the next frontier of growth.

Not the same business

Is it possible to step into the same river twice? The river is in a constant state of change, so is it the same river, and are you really the same? Similarly, does a founder set up a business and then grow it, or is she managing fundamentally different businesses through time. At Notion Capital we take the view that a startup changes so fundamentally on its journey from $1m to $100m in revenue that it’s more akin to a series of different businesses than one business that grows.

“They are not improving the business they have known and loved, they are devising a new business of unknown characteristics”

Stafford Beer

To demonstrate the point, look at man’s progress in traveling at increasing speed. If you were to plot it on a graph you would see the below. No progress for most of human existence. Then explosive growth throughout the 1900s.

The reality is that the increase in speed is driven by completely different technologies. Each with their own sigmoid curve. A new technology is introduced, to start with it is inferior to alternatives, it then rapidly improves, driving progress overall, before diminishing and then being superseded by the next big innovation.

Source: transportgeography.org

What we see here isn’t so different to what’s happening under the surface of a startup’s $1m to $100m journey. A founder adds new capabilities to the business over time, unlocking new growth frontiers. New capabilities take time to build. Hence we shouldn’t be surprised to see times of sluggish growth and then sudden change. In Notion Capital’s Start, Build, Scale framework Stephen Millard describes the key thresholds every startup needs to navigate. Major thresholds exist as a startup tries to find product-market fit, go-to-market fit, and then as it scales and needs to manage rapidly increasing complexity.

Each phase of the startup is fundamentally different. The founder must adapt, the management team will be leveled up, the complexity, scope and focus is different. The things that make a startup successful in one phase will make it unsuccessful in the next, leading directly to its failure. Failing to recognise this. Failing to adapt. Failing to evolve the leadership team. Failing to reconfigure the business to meet the key challenge of each phase is the primary reason startups get stuck.

It’s inaccurate to simply say that a startup grows. I’m not advocating for changing the language we use but it’s simplified terminology that doesn't reflect what’s going on under the surface. Growth, a comparison in revenue generated in period t and t-1 is the result of the system of capabilities built in each period - revenue is an output of and a perfect reflection of the system created. If revenue is very different in the two periods it’s because the system is fundamentally different (or the market has fundamentally changed). This doesn’t mean that many parts of the system aren’t essentially the same, but it does mean that as new capabilities are added the whole is very different to the sum of the parts.

“We can see how the whole becomes not only more than but very different from the sum of the parts.”

Philip Anderson

Sequencing investments

You can start to see how delaying investments into new required capabilities, which could be a mix of technology, people and process, or investing too early in capabilities leads to delays in growth as the startup runs out of steam on its current configuration. 

What can’t so easily be shown in my charts are the dependencies between these changes. There is a natural sequence to adding new capabilities. For example, investing in a sophisticated partnership strategy when the business has not yet achieved a high degree of product-market fit won’t generate additional growth, and could very well achieve the opposite whilst accelerating burn.

Regarding talent, failing to address the need to move from a team of generalists to specialists and then professional leaders of leaders in sales, marketing and customer success will cause growth to stall. Conversely, making these changes too soon will have a similar effect.

Where does this leave us? What if obsessing over growth directly isn’t terribly helpful? If growth is systemic, wouldn’t we be better off focusing on reinventing our little system, and letting growth take care of itself?

Key insights for founders

• Startups don’t grow, increasing revenue is the result of reinvention by adding new capabilities and overcoming thresholds in performance

• Progress towards a threshold can’t be observed in outputs, this can be frustrating and creates uncertainty in whether the strategy is right or execution is flawed

• Underinvestment in key capabilities will delay growth, whilst overinvesting or investing out of sequence will accelerate burn without increasing performance

What drives performance

Tactics are in vogue, Strategy isn’t. Tactics are easy to market; do X and you’ll get Y. Strategy feels less concrete and too idiosyncratic to be usefully discussed. In Phil Rosenzweig’s ‘The Halo Effect’, through rigorous research he dispels the common delusions of business performance concluding that there are only two things we can reliably say impact a company’s performance - Strategy (aka making important decisions) and Execution (following through on those decisions). Other factors may contribute, but it’s hard to say how much, and many are attributes of good performance rather than drivers.

Subscribe now

An important consideration is the difference between drivers of performance vs. attributes of performance, something Rosenzweig demonstrates very clearly:

• Culture: Businesses that grow and are profitable are more fun to work for than businesses that aren’t. But, did ‘culture’ drive that performance or does performance drive employees' assessment of culture?

• People: Great people drive great performance, particularly when in key decision making roles. That’s hard to argue with. But what happens when a business stops performing or the nature of the role changes and individual performance drops? Are they no longer great people?

• Diversity: More successful companies tend to be more diverse. But does diversity drive performance or does performance drive greater diversity?

  ---

The nature of growth

We all know that getting Strategy right matters. But what does it mean for a startup on a growth journey from $1m to $100m in revenue. At Notion Capital we believe that journey is really quite special. It isn’t smooth, there are steep climbs, plateaus and valleys of death. It is a step function, not a linear progression. A well known VC that shall remain nameless published ‘research’ that a startup's growth rate tends to degrade by about 30% per year. That implies a smooth journey from one year of performance to the next. That simply isn’t reflected in what we see and what founders experience.

Working with the Partners at Notion Capital and the Founders of our portfolio companies as they navigate their growth journeys has been invaluable to my understanding of Strategy. It’s been fascinating to combine that insight with two additional sources; i) the works of leading Strategists such as Porter, Helmer, and Rumelt and; ii) insights from complexity scientists and systems thinkers such as Donella Meadows. I’ve boiled that expertise down into seven key insights that reflect a startup’s journey to provide a useful lens through which to view performance.

1. Systemic: A startup is a system - its performance is a perfect reflection of the system it has designed for its environment.

2. Focus and Fit: The market is a selection mechanism; performance is a function of how well the system is designed for the market and challenge at hand.

3. Thresholds: To keep growing a startup needs to overcome thresholds, inflection points at which the nature of the challenge changes and greater fit is needed.

4. Reconfiguration: To overcome thresholds and unlock the next stage of growth a different configuration of people, process, technology and focus is needed.

5. Flexible: The environment is chaotic and new emergent orders are hard to predict. The combination of a strong vision and flexible yet excellent execution is needed to navigate an unpredictable environment.

6. Capital: Startups operate on borrowed time, if they are unable to overcome thresholds with the runway they have, they are unlikely to secure the capital they need to fund new emergent orders.

7. Power: Long term success depends on a startup’s ability to generate power which requires sound strategy and effective execution.

   ---

The nature of systems

A daily battle takes place in the Notion Capital office. We have two sets of colleagues, one group loves the office a little cooler, the other a little warmer. Instead of everyone agreeing to a temperature we can all live with (which through careful experimentation I’ve found to be 22.5℃), we oscillate between temperatures only found in the South Pole and Furnace Creek. One of the key reasons for this is that there are delays between changing the temperature on the thermostat and the impact in the room. If the room gets too warm, the thermostat gets adjusted to 16.0℃, then as the room cools too much someone else changes it to 27.0℃ in an effort to get to a sensible temperature faster. This results in the Notion office experiencing all four seasons multiple times within an eight hour period, constantly overshooting and undershooting that sweet spot of 22.5℃.

The Notion office is a simple system, a startup is much more complex, requires more information to function properly, and experiences more significant delays between analysis, decision, implementation and effect. This leads to large oscillations in under- and over-investment which can be fatal to performance.

“Businessmen don’t know for sure what other businessmen are planning to invest, or what consumers will be willing to buy, or how their products will compete. Their information about these things is incomplete and delayed, and their own responses are delayed. So they systematically under- and overinvest.”

Donella Meadows

Strategy in context

Strategy is context dependent, and so if the system needs to constantly adapt but its adaptation is always delayed, we need a way to get ahead of required investments and put constraints in place that stop over- or underinvestment. I love the now, next, later framework leveraged for product development; we can borrow this approach to help us think through how to sequence key decisions and hopefully more evenly match change and investment with need.

Now - focus

A startup’s focus, its target market and ICP have a big impact on how it needs to design its system of performance. If you don’t have clarity over the challenge you are designing for (which requires focus) and the customers you choose to serve then how can you design a system which fits the challenge?  The greater the challenge the more tightly optimised the system needs to be. If you’ve ever watched the America’s Cup or Formula One you will have seen sailing boats and cars that look alien compared to what’s typical. Both are designed and optimised for very specific and significant challenges. An F1 car is fantastic at overcoming the challenge it was specifically designed for, and completely useless as a functioning car for everyday life.

Where to focus is especially important for early stage startups, but that need to maintain focus and to find new promising areas of focus never goes away. Driving focus, assessing current capabilities and redeploying resources is a permanent need of the now.

“At the core, strategy is about focus, and most complex organizations don't focus their resources. Instead, they pursue multiple goals at once, not concentrating enough resources to achieve a breakthrough in any of them.”

Richard Rumelt

Key questions for founders:

• What are our current resources and capabilities?

• Where should we focus (customer segment, jobs-to-be-done etc.)?

• What should we stop doing, what resources should we free up to re-deploy?

Next - systems

At any point in time there will be multiple limiting factors to growth. Identifying these limiting factors, determining which ones to remove first and matching the level of investment to the need is critical. Exceptional leaders are able to anticipate growth slowing down before it happens, and then make the right investments and right changes to avoid plateaus in performance.

“Whenever one factor ceases to be limiting, growth occurs, and the growth itself changes the relative scarcity of factors until another becomes limiting. To shift attention from the abundant factors to the next potential limiting factor is to gain real understanding of, and control over, the growth process.”

Donella Meadows

Key questions for founders:

• Where are the significant bottlenecks impacting our performance?

• What is the right sequence of changes/investments?

• How good do we really need to be to unlock growth?

Later - power

Eight years working for a strategy consultancy razor focused on driving free cash flow has made me more conservative than the typical venture capitalist. I really care about efficiency and whether a business can drive healthy margins in the long run beyond just paper valuations. The single most important factor for longer term success is the ability to generate market power - the power to grow share and charge prices that funds future innovation and investments. 

Power generally takes time to emerge and depends on the structure of the industry, competitive dynamics, technological change etc. This necessitates a longer term perspective and vision from founders to help guide short and medium term decision making,  ensuring that they move the business towards a position of power rather than away from it.

“A body of theory which links firm characteristics to market outcomes must provide the foundation for any fully dynamic theory of strategy. Otherwise dynamic processes that result in superior performance cannot be discriminated from those that create market positions or company skills that are worthless.”

Hamilton Helmer

Key questions for founders:

• How does the structure of our industry shape our ability to develop power?

• What are our potential sources of power and when and how are they likely to emerge?

• What changes are we anticipating that could have a profound impact on the industry and/or our position?

  ---

References and recommended reading:

• The Halo Effect by Phil Rosenzweig

• Good Strategy Bad Strategy by Richard Rumelt

• 7 Powers by Hamilton Helmer

• Thinking in Systems by Donella Meadows