Finding Product-Market Fit

This is a narrative theme with a typological spine, synthesising how practitioners in this wiki — Rahul Vohra, Jag Duggal, Sean Ellis, Todd Jackson, Bob Moesta, April Dunford, Christian Idiodi, Uri Levine, Raaz Herzberg, Kunal Shah, Mike Maples Jr., and Dalton Caldwell — disagree productively about how to detect and force product-market fit. The central argument: PMF is not a threshold you plan for but a state you discover through disciplined search, and the disagreement worth tracking is between demand-side signal methods (measure how users would feel if the product disappeared) and supply-side structural methods (position correctly so the right users find the product at all).

The core claim

PMF is not a milestone on a roadmap — it is what happens when a product’s value proposition lands squarely enough on a genuine struggling moment that users return, tell others, and resist leaving. Every practitioner in this wiki agrees on that much. The productive disagreements are about direction of travel: do you measure your way there, or reason your way there from first principles?

The measurement camp (Ellis, Vohra, Duggal) starts from user signal: survey active users, segment ruthlessly, let the data tell you which sub-segment already loves the product, and build toward that love. The structural camp (Moesta, Dunford, Idiodi) starts from causal understanding: find the struggling moment before you build anything, position against real competitive alternatives, earn reference customers whose public endorsement proves value. Uri Levine cuts across both: PMF is retention, full stop — every other metric is a proxy.

Before you start: the wrong ideas trap

The search for PMF begins earlier than most founders think — with idea selection. Dalton Caldwell describes the tarpit: an idea that generates easy positive feedback, has been attempted by many founders over many years, and rarely succeeds. The danger is precisely the validation. Social-coordination apps have attracted enthusiastic user interviews since the 1990s; that enthusiasm has never produced durable businesses. The diagnostic question is not ‘do people say they want this?’ but ‘can I explain specifically why every prior attempt failed in a way that is unique to me?’

Mike Maples Jr. frames the precondition differently: a breakout company requires an inflection — a step-change in what is technically, legally, or socially possible — and a founder whose background positions them to exploit it before consensus forms. The Inflections and Pattern Breakers framework suggests that founder-future fit precedes product-market fit: the right idea, at the right moment, for the right person. PMF cannot be discovered if the search begins in the wrong territory.

The demand-side measurement method

Sean Ellis invented the survey question that anchors half the practices in this wiki: ‘How would you feel if you could no longer use this product?’ Forty per cent or more answering ‘very disappointed’ is the threshold he identified empirically across early Silicon Valley companies. The insight is the framing inversion — asking about loss, not satisfaction, because loss aversion produces more honest signal than satisfaction questions.

Rahul Vohra operationalised this into the Product Market Fit Engine: measure with the Ellis question; segment the ‘somewhat disappointed’ group by whether the product’s core benefit resonates; build a roadmap from what fans love most and what the ‘almost convinced’ group still lacks. The algorithm is self-reinforcing — deepening the product for fans and addressing the specific objections of the near-converts both raise the score simultaneously. Crucially, the ‘not disappointed’ group is ignored entirely; their feedback points away from PMF, not toward it.

Jag Duggal at Nubank raised the bar to 50% to correct for Brazilian cultural optimism in survey responses, and converted the score from a diagnostic into a hard gate: products are not scaled until they clear it. His Bullseye Cohort method solves the borderline case — when the overall score is 40% but a sub-segment scores 70%, the job is to reverse-engineer what that cohort shares and expand it. The Payments Assistant went from a borderline score and hundreds of thousands of monthly actives to over ten million after fifteen months of iteration targeted at users with four or more billing commitments across two or more payment rails. The bullseye method makes PMF iteration concrete: find who already loves it, understand why, make more of the base look like them.

The supply-side structural method

Bob Moesta and April Dunford share a prior to measurement: you cannot measure your way to PMF if you have not first understood the causal structure of demand. Moesta’s Jobs to Be Done framework centres on the struggling moment — the specific context-and-outcome vector that causes someone to act. Struggling moments create demand; product launches do not. A product positioned against the wrong struggling moment will not be hired regardless of how well it executes, because users will not recognise it as the answer to their problem.

The practical implication: the right research is not a satisfaction survey but a causal interview with people who recently switched. ‘Bitchin’ ain’t switchin” — complaints do not predict behaviour change. Only past behaviour does. The four-forces model makes switching visible: Push (current situation untenable) and Pull (attraction to a new outcome) must together exceed Anxiety of the new and Habit of the present. Adding features increases Pull but also Anxiety; reducing switching friction is often cheaper and faster than building more product.

April Dunford’s Product Positioning framework starts one step earlier still: before asking whether users love the product, ask what they are comparing it to. The competitive alternative is almost always status quo — ‘doing nothing’ wins roughly 40% of B2B deals. A product positioned against the wrong alternative will be evaluated against the wrong criteria and lose deals it could have won. Positioning is the five-step sequence: competitive alternatives → differentiated capabilities → the value those capabilities enable → the customers who care most about that value → the market category that makes the value obvious to them. Category is chosen last, not first.

Christian Idiodi’s Reference Customer method converts this structural analysis into a PMF definition: PMF is reached when a threshold of target customers are willing to publicly stake their reputation on the product. For B2B, six to eight; for B2C, fifteen to twenty-five. Willingness to recommend publicly costs the person something — their reputation — which is what makes it an honest signal. A customer who enthuses privately but hesitates before leaving a five-star review has revealed an unresolved problem.

When signals differ: detection in practice

Raaz Herzberg‘s Wiz account provides the clearest real-world illustration of what PMF detection looks like across the methodological divide. Before fit, prospects said ‘sounds interesting.’ After the pivot to cloud security, the question changed: ‘how do you price this, when can we start a proof of value?’ No survey was needed. The quality of inbound questions shifted in a way that was unmistakable. Herzberg’s framing aligns with Uri Levine‘s: retention is the proof, and the qualitative shift in engagement is what retention looks like before you have enough users to measure it.

Todd Jackson’s PMF Levels framework (First Round Capital) makes the spectrum explicit. Sixty per cent of B2B companies never pass Level 2 — developing PMF — because they mistake an important problem for an urgent one, or a ‘nice to have’ for a ‘must have.’ His four-Ps pivot model identifies which dimension is misaligned (Persona, Problem, Promise, or Product) and argues that most founders pivot too shallowly: a ten per cent adjustment when a two-hundred per cent one is needed. Lattice kept the persona and changed the other three; Ironclad changed only the promise (repositioning into an existing category) and opened the demand floodgates.

Eric Ries offers the practitioner’s heuristic that cuts through all methodologies: if you are asking whether you have PMF, you do not have it. PMF is unmistakable. The covert signal from the measurement methods and the structural methods converges on the same place — users who find the product essential behave differently from users who find it useful.

The efficiency threshold

Kunal Shah‘s Delta 4 Framework introduces a necessary condition that none of the measurement or structural methods make explicit: PMF requires an efficiency gap of at least four points on a ten-point scale against the existing solution. Below a Delta 4, behaviour change is reversible — users return to the old way when the new product stumbles. Above it, three things follow automatically: the behaviour change becomes irreversible, users tolerate failures, and word-of-mouth becomes structural rather than engineered. The Uber-versus-traditional-taxis gap was approximately six points; online suit-buying versus in-store was negative, which is why it has never taken hold despite decades of effort. Delta 4 explains why PMF is harder to reach than measurement alone suggests: a product can score well on the Ellis survey while still sitting below the efficiency threshold needed to drive organic growth.

The scope of disagreement

The demand-side and supply-side camps are not incompatible — they describe the same state from different directions. A product that has reached the Ellis threshold has almost always found the right struggling moment and reduced switching friction; a product with strong reference customers has almost always reached or is approaching the Ellis threshold. The disagreement is about where to start.

The measurement-first approach (start surveying as soon as you have active users, let the data drive the roadmap) works well when you have enough users to segment and when the core struggling moment is already roughly correct. The structure-first approach (find the struggling moment, earn reference customers, only then measure) works well when the user base is too small to survey or when the core struggling moment has not yet been correctly identified.

Dalton Caldwell’s pre-PMF warning is the clause that applies to both: analytics loops and growth hacking are a waste of time before you have genuine user love. Twenty to thirty per cent of founder calendar should go to customer meetings; the Collison Install (showing up in person to implement alongside the customer when they go quiet) is more diagnostic than any A/B test. The search for PMF is an immersive empirical process, not a quantitative optimisation — until the signal is strong enough to optimise.