The Spreadsheet That Lies to You

Picture the slide every seed-stage investor has seen a thousand times: a total addressable market number in the billions, divided by the number of competitors, multiplied by a modest-sounding 10%. The math resolves to a clean, confidence-inspiring figure. The founder nods. The deck moves forward.

The number is a fiction.

Consider a founder who knows the kitchen appliance market cold — supply chains, manufacturing costs, distribution channels across Spain — but has never baked a loaf of bread or stretched a pizza dough. He calculates that capturing 10% of the country’s professional bakers, pastry chefs, and pizzerias would make him a billionaire, so he builds a more efficient oven and waits for orders. The pitch is airtight on paper: want to work more efficiently? Buy our oven. What the spreadsheet cannot capture is that professional bakers do not buy ovens the way consumers buy toasters. They buy tools that fit their specific heat curves, their dough hydration levels, their production rhythms. A founder who has never baked cannot ask those questions, because he does not know they exist.

This is the infrastructure-without-craft trap. The founder understands the platform — the oven, the app store, the distribution network — but not the practitioner’s lived problem. Market share gets modeled as something you arithmetically claim from a static pool, not something you earn by solving a specific person’s specific frustration better than anyone else.

Venture capitalists have a name for this approach: top-down market sizing. It starts with a macro number — the global SaaS market, the European foodservice equipment sector — then works downward by percentage. Most experienced investors treat it as a red flag for exactly this reason: it measures the ceiling of a market without interrogating the floor of actual customer demand. Bottom-up sizing forces founders to count real buyers, real use cases, and real willingness to pay. Top-down sizing lets founders avoid that reckoning entirely while appearing rigorous.

The spreadsheet does not lie by accident. It lies because percentages compress uncertainty into something that looks like strategy. Ten percent of a billion is a hundred million. That sentence contains no information about whether a single customer will ever switch.

Domain Knowledge vs. Industry Knowledge: A Critical Distinction

Knowing an industry and knowing a craft are two entirely different things — and conflating them is where B2B and prosumer startups quietly bleed out.

The oven founder in the parable knows the kitchen appliance market cold. He can cite supplier margins, competitor SKUs, and distribution channel economics across Spain. What he cannot do is bake bread. That gap — between market intelligence and domain expertise — means he has no felt sense of how a baker actually works. He doesn’t know that deck temperature uniformity matters more than peak heat output, that steam injection timing is non-negotiable for crust development, or that a baker’s workflow is built around the oven’s rhythm, not the other way around. His “more efficient oven” is optimized for a workflow he has never lived inside.

This is the domain knowledge deficit, and it produces a specific failure mode: features that look rational on a product roadmap but collapse the moment a real practitioner touches them. The founder builds what efficiency looks like from the outside. The baker needs what efficiency feels like from within a six-hour production cycle.

Founders who cannot use their own product are uniquely exposed here. They rely on user interviews and market research to approximate craft knowledge — inputs that are always incomplete and frequently misleading, because practitioners struggle to articulate tacit knowledge. A pastry chef won’t tell you she needs 40 seconds of residual heat after the door opens. She just knows the oven that doesn’t have it ruins her choux.

The AI era has made this problem structurally worse. Rapid prototyping tools, no-code platforms, and generative AI let a small team ship a polished, functional product in weeks. Speed used to be a forcing function for focus. Now it removes the friction that once slowed founders down long enough to discover what they didn’t know. The result is a wave of technically sophisticated products — clean interfaces, solid architecture, credible demos — with hollow utility at the workflow level. Product-market fit research starts with the question of whether a product solves a real problem. In craft-dependent verticals, you cannot answer that question from the outside.

What ‘Half-Baked’ Really Means in 2025

A half-baked product in 2025 rarely crashes on launch. It loads fast, renders cleanly, and survives a 30-minute demo without a single error message. The problem runs deeper: the product is technically sound but professionally useless — built around how its creators imagined work happens, not how it actually does.

This distinction matters more now than at any point in the previous decade. The AI product wave has flooded the market with tools that generate genuine awe in controlled settings. A legal research assistant that summarizes case law in seconds looks transformative when a founder demos it to investors. It looks different when a practicing attorney tries to rely on it across a 60-hour case with non-negotiable accuracy standards. The gap between those two moments — between impressive and indispensable — is where most AI startups quietly die.

The oven founder in the parable understands market size. He does not understand baking. That gap is not a personality flaw; it is a structural blindspot that no amount of market research from the outside fixes. Founders who have never worked inside the professions they are disrupting build products that model the workflow they assumed existed. Professionals then encounter a tool optimized for a job description nobody actually holds.

What makes this failure mode so persistent is that it passes every early filter. Seed investors see a clean interface and a compelling narrative about efficiency gains. Design partners give positive feedback during structured pilots where the product is set up to succeed. Early retention numbers look acceptable because the first cohort is composed of early adopters willing to work around limitations. By the time real-world usage patterns expose the misalignment — the tool doesn’t fit into existing systems, the output requires too much human correction, the workflow integration was never fully thought through — the company has already built its roadmap around the wrong assumptions.

Product-market fit diagnostics need to catch misalignment at the workflow level, not just the feature level. A product that professionals tolerate is not the same as one they trust. In 2025, that line separates companies with durable retention from companies with a great deck and a slow bleed.

The Missing Step: Talking to the Bakers Before Building the Oven

The oven founder analyzed Spain’s entire commercial kitchen market and never once watched a baker work a morning shift. That gap — between research about users and time spent with users — is where product-market fit goes to die.

Proper customer discovery is not a 30-minute Zoom call about pain points. It means standing in a commercial kitchen at 5 a.m., watching how a pastry chef sequences her prep, where she hovers near the oven door, when she overrides the temperature dial, and what she mutters under her breath when a batch comes out wrong. The workflow contains the product brief. The interview rarely does.

The Spain market analysis is a precise model for how tech founders confuse market research with user research. Counting the number of pizza makers, pastry chefs, and bakers in a country tells you a population exists. It tells you nothing about purchasing authority, equipment switching costs, calibration dependencies, or why a baker who has run the same deck oven for eleven years trusts its hot spots like a co-worker. Those are the variables that determine whether a new product gets adopted or ignored — and no spreadsheet generates them.

One honest conversation with a working baker would surface questions no TAM calculation thinks to ask. Does the kitchen lease specify approved equipment? Who actually signs the purchase order — the head chef or the operations manager? What happens to recipes calibrated to the old oven’s temperature variance when you swap in something more “efficient”? Efficiency is not a universal value in professional kitchens; consistency is. A founder who spent two weeks doing contextual inquiry inside real bakeries would know that before writing a single line of product requirements.

The failure is not ignorance — it is the substitution of legible data for lived context. Market sizing is legible. Workflow observation is slow, inconvenient, and resistant to being dropped into a deck. Founders choose the spreadsheet because it produces confidence quickly. Real user research produces questions, and questions feel like delay. They are not. They are the only reliable early-warning system for whether a product solves a problem users will actually pay to fix.

Why Investors Keep Funding This Pattern Anyway

Early-stage investors are pattern matchers operating under time pressure. A founder who walks in with a billion-dollar TAM, a clean market segmentation slide, and the quiet confidence that comes from having done the spreadsheet work hits almost every heuristic a seed or Series A investor has been trained to reward. Large addressable market. Credible founder. Differentiated technology. Simple value proposition. Check, check, check, check.

The venture capital model makes this worse in a structural way. A fund that returns 3x needs its winners to be enormous. That math pushes investors toward swinging at large markets even when the product thesis underneath the market size hasn’t been stress-tested. A story where the startup captures 10% of a $10 billion market is fundable on its face — the outcome pencils out, the risk feels diversifiable across a portfolio, and the downside is capped at the check size. What that story doesn’t require is proof that any real customer has a burning need for this specific product at this specific price point.

Accountability for product-market fit arrives at the wrong moment in the funding cycle. By the time an oven actually ships — by the time a pizza maker or pastry chef puts the product through a real production run and reveals whether the efficiency gains are real — a seed round and likely a Series A have already closed. The investors who wrote those checks made their decisions based on a deck, a demo, and a market analysis. The founder who raised on a “we only need 10% of the market” thesis has already spent twelve to eighteen months building to spec without ever confirming that the spec matches what operators actually need.

This timing gap is the mechanism that keeps the pattern alive. Startup funding decisions get made on market narratives. Product-market fit validation happens after capital is deployed. The incentive to ask hard questions about whether a professional baker would actually switch ovens — and why, and at what cost, and against what switching friction — arrives too late to change the outcome. By then, the round is closed and the build is underway.

What Founders (and Readers) Should Take Away

The oven founder’s mistake wasn’t ambition — it was ignorance worn as confidence. The antidote is earned expertise. Build with practitioners, not just for them. Spend time inside the bakery before you spec the oven. If you can’t describe exactly how a pastry chef loads trays at 5 a.m. or why a pizza maker rejects a particular heat distribution pattern, you don’t understand the problem well enough to solve it.

That practitioner knowledge is what bottom-up validation actually tests. Forget the top-down market math. Spain’s commercial baking sector, your total addressable market, your 10% assumption — none of that tells you whether a single real customer will pay. Replace the spreadsheet exercise with one brutal question: can you name ten specific people who would write a check today, and do you know precisely why? Not a demographic. Not a persona. Ten named humans with a specific problem that your product solves better than anything else available to them right now. If you can’t answer that, product-market fit is not validated — it’s theorized.

The AI era makes this discipline more urgent, not less. When the cost of building software drops toward zero — and it has, dramatically, for any team using today’s code-generation tools — the ability to ship a prototype no longer separates winners from losers. Every competitor can spin up an MVP in weeks. What they can’t replicate quickly is deep, domain-specific understanding of a real workflow. That understanding — knowing why the artisan baker rejects your oven before your engineer finishes the demo — is now the primary source of sustainable competitive advantage in startup product development.

Founders who grasp this shift stop treating customer discovery as a pre-launch checkbox and start treating it as the core technical skill of the company. They embed in the industry. They hire people who have done the job, not just studied it. They measure product-market fit not by market-size estimates but by retention curves, referral behavior, and the willingness of early users to feel genuine pain when the product is taken away. That signal — specific, human, unambiguous — is what separates a real business from a half-baked one.