EP407 · Economy · first published 2026-09-10
Hard-Won Lessons in Investing | Petri Lehmuskoski | Negotiator 407
Petri Lehmuskoski, founder of Gorilla Capital, explains why product market fit is an unusable term: it has never been defined, so it cannot be measured. In its place his book More Scars Than Trophies offers four measurable proof stacks, the first two being buyer proof and value proof — and value is only counted after the customer's own cost, at worst the decision-maker's reputational cost, has been subtracted. Running through it is the concept of hull speed: more money does not raise the speed if the hull will not take it, and a company's hull speed is the product of the entrepreneurs in it. Lehmuskoski explains why revenue is a poor measure of value, why growth and scaling are not the same thing, why an exit has to be built from the day the company is founded, and how AI consumes hard-won evidence ever faster. It closes on the camel and the unicorn as investment strategies. Published 10 September 2026.
Hard-Won Lessons in Investing | Petri Lehmuskoski
Summary: In episode 407 Sami Miettinen interviews Petri Lehmuskoski, founder of Gorilla Capital, about his book More Scars Than Trophies. Its central claim is that product market fit is an unusable term, because it has never been defined and therefore cannot be measured — and in its place Lehmuskoski offers four measurable proof stacks. Published 10 September 2026.
Why product market fit does not work
Lehmuskoski says the book began when he gathered fifteen years of his own writing into one place and fitted it to the AI era. What emerged was simple: product market fit and problem solution fit are very weakly defined. They are terms, but what they contain and how they form has never been described. Everyone has their own understanding, which makes it hard to say what stage anyone is actually at.
The practical problem recurs throughout the episode. When a founder says they are “close to product market fit”, it tells an investor nothing — and the early end and the late end of the same concept are two entirely different worlds. Lehmuskoski’s solution is to break the concept down to the molecular level so that it becomes measurable.
Four proof stacks
The book divides a company’s journey into four proof stacks, or gates. The last two — the scaling gate and the exit gate — resemble conventional thinking. The first two are where Lehmuskoski sets things at a fundamentally different angle.
Buyer proof. Somebody has to buy something, preferably in a market transaction — barter, the public sector and charity are alternatives, but the honest signal is somebody putting money on the table. It can be money, resources or significant time, as long as it is a measurable commitment. “I’m interested” is nothing but the expression of an opinion, and it does not convert into a sale.
The term is deliberately buyer proof, not user proof, and the distinction is the single most useful one in the episode: a user does not make an economic decision. A user can be as enthusiastic as you like, but if no money moves through them, it proves nothing about the business.
Miettinen connects this to an observation from his episode with Mika D. Rubanovitsch: the purchase funnel is often theatre, free consulting in which an enthusiastic salesperson talks to an enthusiastic non-decision-maker. Both can enjoy themselves and both can spend a great deal of time on it. A good salesperson eventually asks whether the other party has the authority to buy, and if not, whether they will bring the person who does to the next meeting. Lehmuskoski’s addition is temporal: a startup has limited time, so if you do a lot of theatre, there is no time left for actual trade.
Value proof. This is the book’s most original move, because it looks in both directions. Value is not only what the customer gets, but the net value after the customer’s own cost has been subtracted. That cost may be training, the cost of switching systems, an investment — or, at worst, a reputational cost: whoever makes the decision can lose their standing because of it, and for many that is a large cost indeed.
Lehmuskoski’s observation is that this is rarely considered: we see only what the buyer pays us, and assume the benefit exists. Yet understanding it is precisely what tells you whether you can raise the price or are already too expensive.
The extreme example is an ERP switch: if a company runs Microsoft Dynamics and has to move to SAP, it is like a heart transplant in flight, because the whole operation is built around the ERP. There are projects, consultants, habits, possibly two systems running in parallel and double costs — and the benefit has to be correspondingly larger. This explains why large system projects fail: nobody worked out what costs arise on the side. The other extreme is SaaS at its best: everything runs, you put an email in and it goes, and the experience has been honed with a thousand customers at marginal cost. Even there friction remains — HubSpot to Salesforce, manual to systematised, pen to keyboard; all of it carries learning friction.
Measurement turns development into a loop
Once something is measurable, you can see which way it is developing. Lehmuskoski stresses that the model is not a straight march forward but a loop: perhaps this does not produce enough value, perhaps the buyer is wrong, and you move it systematically towards a more optimal state. Buyer proof and value proof alternate: first the customer parts with money, then in the next cycle you ask whether that same customer would pay five hundred instead of one hundred. It is not only upsell — it is about not charging too little.
And the evidence decays. The market moves much faster than it used to, so what you know today may not be valid in two years.
Autovex is his example of genuine pain: people had a concrete problem — a wreck sitting in the yard — and the alternative was listing it online, waiting for offers, or spending days driving from dealer to dealer. A simple but clever idea that scaled, and eventually a foreign acquirer paid serious money. Miettinen asks directly whether this is typical or one in a million; Lehmuskoski’s answer is that it is the target state.
Revenue is a poor measure of value
One of the sharpest points in the episode. Revenue does not work as a measure of value proof, because revenue arises even without value proof: with a tough, good salesperson or founder you will generate revenue — but is it repeatable?
Lehmuskoski has a live example: a company where value proof clearly exists (current customers pay more each time) but buyer proof does not. What is needed is more understanding of what kind of customer buys and what has to be present in that customer for the decision to happen.
Growth and scaling are not the same thing, and in his view they are constantly confused. In growth you add inputs and sales follow from the added inputs: inputs and outputs grow at the same rate. In scaling you can reduce inputs and still generate more sales — inputs and outputs diverge. The scaling gate is therefore an efficiency measure, and there too you have to measure whether you are going the right way, because the typical mistake is doing something for a very long time and only then noticing the direction was wrong.
Hull speed
The book’s best-known metaphor is hull speed, and the Naantali sailor explains it precisely. A displacement hull — not a planing one — has a mathematical top speed the hull will not allow you to exceed. Fit whatever engine you like, whatever propeller: water flies, the stern drops, fuel burns, and the speed does not rise.
Every entrepreneur and every company has its own hull speed, and a company’s hull speed is the product of the entrepreneurs in it — a product in the sense that the factors multiply rather than add. Investors affect it too. The typical mistake is trying to solve it with more funding: take two or three million into the company, go bow-wave first, and nothing actually happens. Miettinen’s image is three turbo engines on a sluggish-hulled boat. Lehmuskoski’s correction is sharp: it is not merely wasteful but dangerous — you burn money and go the wrong way for so long that you then have to reverse.
The classic version is hiring salespeople without evidence: hire without knowing what they say, conclude they achieve nothing, fire them, hire the next lot, and by the third or fourth round wonder why it does not work — until you go out and sell yourself and find the money is gone.
Hull speed also includes the founding team’s chemistry and how the organisation handles new people who are not founders. A founder can raise hull speed and can also lower it.
Right decisions, too early
Asked why startups make right decisions too early, Lehmuskoski gives two reasons. First, perspective: many advisers and investors are corporate people and advise from that perspective, so late-stage advice arrives far too early. Second, there is no understanding of what the startup is currently trying to establish — so it scales without knowing who the customer is.
“Stage match proof before irreversible decisions” means in practice a simple question: have we been able to prove that customer behaviour is repeatable before we hire those three salespeople?
This does not mean demanding perfection. Nothing can be a hundred per cent certain, so risks must be taken — but as conscious bets. When 50 to 70 per cent is clear and you believe the rest will hold, you have to take the risk; in a competitive environment you must. The difference is not walking blind into a gamble.
Customer interviews are not enough for the same reason “I’m interested” is not enough: they must involve something the other party gives up. The version demonstrated on Miettinen’s own podcast: it is easy to say the podcast is superb, but would you pay fifty euros for it?
An exit does not happen by accident
The fourth proof stack is the exit, and Lehmuskoski is blunt: the driftwood theory of exits is a bad theory. An exit has to be built and measured — are we going that way? — rather than hoped for, as if the phone will ring by itself.
When should you start thinking about it? Gorilla’s position: when the company is founded. The reason is structural rather than romantic — raising funding affects everything through to exit, and from the first round onwards doors and exit options start closing. The focus must be on building a good business, because nobody buys a bad one, but the exit should be there alongside it the whole time.
As a counterweight, Lehmuskoski warns against the opposite extreme: obsessively building a company for one specific dream acquirer is dangerous, because after the first contact it may turn out they are not interested after all. Miettinen adds the AI-era curse: you hand over the DD material, they say it is not for them — and build the competing product.
Useful or necessary
Nice to have or need to have. A useful company produces added value but is replaceable, and AI replaces the merely useful very easily. The necessary kind contains a deeper insight, and finding that from outside is far harder.
This is the book’s finest single idea about competitive advantage: technically a product can be copied, but there may be a particular step, a particular order, that adds a great deal of value for some customer group. From outside it looks illogical — why on earth does it behave like that — and that is exactly why it is hard to spot from the product alone.
The proof stack as a board tool
Lehmuskoski has built three tool sets on top of the book, one from the founder’s perspective and one from the board’s. The point is that everyone talks about the same thing at the same time. Boards and investors are usually very smart people, but the dilemma is that founders and they talk past one another or from different stages. From the founder’s side it shows up as advice so varied you cannot choose between it.
The risk is micromanagement. Lehmuskoski’s boundary comes from experience: in eighteen years of angel investing, situations where the investor imagines they are the entrepreneur without being there all day have produced very poor results. Steering from a distance is like pushing on a rope; the founder makes the decisions, and that is precisely why you invest in founders.
Miettinen’s question about boards versus owner steering produces the same answer from another direction: when an owner has money, power and energy, the board easily becomes a redundant layer in between. The tool set has a section for managing exactly that. The discussion runs through Ville Tolvanen’s “board as a service” model — a board that helps within hull-speed limits and does not dump administrative burden on the CEO — and arrives at the startup’s annual calendar: a corporation’s annual cycle looks much further ahead, but the thinking still applies.
From which comes the book in one line: you get what you measure — and if you measure nothing, usually nothing happens.
The angel investor’s view
When there is no evidence yet, the evidence is the founder. An angel invests in the founder, not the market and not the product — and specifically in whether this person appears capable of raising their own hull speed.
When does money help? Money has to have objectives and has to be moving towards them. Lehmuskoski remembers from his own first rounds that the money simply disappeared. If you raise too much and the objectives are too far away, it does not move in sync.
What AI does to the evidence problem
This is the episode’s most current section, and it runs three ways.
Evidence decays faster. Scaling proof is challenged by the constant arrival of new tools: if a salesperson-led approach worked, AI has brought new tools and new prospecting methods, and you have to go and establish it again.
Bad evidence can be made to look beautiful. Decks and presentation material have jumped a light year forward, and it is now hard to say what is fact, what is invented, what is assumed and what is merely wished for. Lehmuskoski reports that Americans have written about decks arriving with AI prompt-injection attempts embedded — instructions about how an AI should respond when evaluating them.
The gate has disappeared. Being able to build software with a team of coders used to be a substantial gate not everyone could pass. Now even a vibe coder produces minimum viable products over a weekend. This creates a peculiar risk for startups: if you bring the customer too much value, the customer acquires an interest in replacing you with their own product. Lehmuskoski notes that whether they succeed is another matter — it is always a project of its own and has its price — and that an internal build does not necessarily scale anywhere: the sales director does not feel the benefit, and administration suddenly carries a great deal of new cost.
What must not be outsourced. The search for buyer proof and value proof, because these typically begin from weak signals — from the questions you did not know to ask. The typical outsourcing is a salesperson talking to customers, but a salesperson only asks what they know to ask. A customer may drop a small weak signal and the salesperson lets it pass because it means nothing to them — even though it might reveal who actually decides and whose budget is in play. The further up the organisation you sell, the larger the budgets being controlled.
This connects to Lehmuskoski’s answer on where humans still beat AI: you can use AI for the measuring, but in identifying and finding buyer proof and value proof the human is still far better. The reason is that communication is two-way — it is not what you know to ask, but what arrives as an aside, and AI does not necessarily recognise that it might matter.
Unfair advantage, and an investment banker’s real pain point
Unfair competitive advantage often comes from the founder having been the buyer of that problem themselves: I solved a problem I wrestled with daily, and I talked to others with the same problem. Then levels one and two of buyer proof and value proof do not have to be established separately. The bonus is that as a buyer you already looked for solutions and did not find them, and you were the target of everything that was pushed at you and did not work.
Miettinen turns this onto his own field, and it produces the episode’s sharpest concrete example of missing value proof. An investment banker’s pain point is not building a data room, an info memo or a teaser, nor distributing them — there is an endless supply of vendors with a better pitchbook, generator, profile builder or a perfect data-room kit. The pain point is negotiating the right agreement that produces a closing, and specifically during the exclusivity period. Lehmuskoski’s diagnosis is precise: buyer proof may exist, but the value proof is very weak. From outside something can look terribly hard while having almost no effect on the deal — the real pain is somewhere else entirely. Lehmuskoski describes Gorilla’s own Exitension tool, which hands an M&A adviser a ready data set: it is built from the seller’s end, and whether it carries enough value is another question.
Toptronics, Grand Theft Auto and raising the price
The episode’s most entertaining thread is Lehmuskoski’s background in games since 1983. Toptronics was Finland’s leading games importer and dealer back when computer games were sold as physical deliveries, on cassette. He has been involved in selling every Grand Theft Auto except the first and the forthcoming sixth.
The story returns to the book’s framework: Grand Theft Auto should have been priced at 120 rather than 80, because buyer proof had been totally established and value proof was there. With the fourth instalment the price was raised radically and sales volumes only rose. Miettinen confesses to having been one of Tampere’s better crackers — meaning it all flowed past the buyer proof for free.
The camel and the unicorn
Towards the end the conversation returns to Gorilla Capital’s philosophy, which Miettinen describes as an index fund for angel investing. The third fund holds over 70 investment decisions across 70 different startups. The rationale is diversification: at the early stage, picking the winner has become far harder than ever before, and diversification solves part of that problem.
Comparing the camel and the unicorn is, Lehmuskoski says, a little dangerous, because they are not the same kind of thing: the camel is a way of getting somewhere; the unicorn is an outcome. A camel can become a unicorn — you just do not go via the extreme, you do not burn millions and then bet on whether anything came of it, but proceed systematically and tighten the pace as the scaling proof holds.
The difference shows in the structure of returns. Gorilla’s investors see a return from a much smaller exit, because it does not have to be tens of millions, and that is why one of the most important measures at the investment stage is valuation: it has to be realistic so a return is possible from an ordinary exit rather than only a unicorn exit. Miettinen’s contrast is the portfolio where a handful of investments must cover the whole fund and the rest can be allowed to die.
Scars and trophies
The book’s title comes from the fact that every chapter carries a wound from Lehmuskoski’s own life — mostly mistakes, but also positive wounds and successes. A couple of stories run through the whole book because they involve both a wound and, at some point, a large success; one of them is the Xbox story, which begins with another console entirely and moves through a big rise, a failure, another rise and then a radical decision. In every case he was personally involved as entrepreneur, investor or adviser.
The closing lines say the same thing as the decay of the proof stacks: nothing achieved is permanent. “I always have to reinvent myself to stay competitive in this competitive environment.”
A note on the source
The episode was transcribed with MacWhisper and the transcript carries no speaker labels. With a single guest the speakers are reliably distinguishable. There are no gaps in the transcription. The book’s English terms (buyer proof, value proof, scaling proof, proof stack) appear in the transcript in numerous mangled forms and have been normalised to the book’s own usage in the cleaned cue file. Spoken names and expressions that could not be verified have not been guessed.
Watch
The recording lives on the Neuvottelija channel: Karvaat opit sijoittamisesta | Petri Lehmuskoski | Neuvottelija 407.