Why Most Market Sizing Estimates Are Wrong by 20% or More

Every pitch deck and strategy memo eventually leans on a market size number, and almost nobody stress-tests where it came from. TAM figures get copied from slide to slide, quietly ageing past their original assumptions, mismatched against shifting industry classifications, or inflated to make an opportunity look irresistible to investors. Epignosis Insights' review of government classification data, industry-association benchmarking, and recent IPO filings shows the errors are rarely random rounding; they follow predictable patterns, and once those patterns are known, a bad number can be caught before it becomes a bad decision.

The Classification Problem Almost No One Accounts For

Most market sizing work still starts with a NAICS code pulled from U.S. Census Bureau data, but the classification system itself keeps moving. The Census Bureau's 2022 NAICS revision altered more than 100 of the 1,057 six-digit industry codes in use, consolidating some categories and splitting others, which means a market defined against the prior code set no longer lines up cleanly with current government data. The Economic Census also excludes most government-dominated activity and portions of the services sector by design, so the base figures analysts build from are incomplete before a single segment filter is even applied. A market size inherits every blind spot of the classification system underneath it, and those blind spots shift on a schedule most sizing exercises never check.

When the Industry That Sizes Markets Can't Size Itself

Esomar, the trade association that has tracked the global market research industry since 1947, is candid about its own limits: defining and sizing the insights industry, in its own words, has become a moving target as the boundaries between traditional research, software platforms, and data analytics continue to blur. Research software now accounts for roughly 41 percent of global insights industry turnover against 36 percent for traditional market research, a split that would have looked reversed only a few years earlier. If the professional body that measures markets for a living cannot hold its own category boundaries steady year over year, a boundary drawn once for a single company's TAM deserves the same scepticism.

The Trillion-Dollar Habit of Rounding Up

Company filings themselves are a recurring source of the error. SpaceX's 2026 IPO prospectus put its total addressable market at $28.5 trillion, attributing $26.5 trillion of that figure to AI opportunities and describing the company's reach as the largest actionable market in human history. Reporting on Anthropic's subsequent IPO preparations put its pitched TAM at over $30 trillion, surpassing SpaceX's figure within months, with NYU Stern finance professor Aswath Damodaran telling the Wall Street Journal, as cited in that reporting, that SpaceX's AI market estimate was already reaching the end of what's plausible and pushing beyond it. When back-to-back IPO filings pitch addressable markets several multiples larger than any plausible subsegment of global GDP could support, the figure stops functioning as an estimate and starts functioning as a narrative device, which is exactly why an independently reconstructed number matters before pricing or planning decisions lean on someone else's TAM.

The Execution Gap Hiding Behind the Estimate

Bain & Company's 2026 B2B Growth Agenda survey of more than 1,100 senior executives across 18 industries found that 42 percent missed their 2025 revenue growth targets, up from 32 percent the year before, even though 86 percent of those same leaders had expected to hit their goals; 91 percent are again confident about 2026. A widening miss rate like that is not purely an execution failure. It is evidence that the market-size assumptions sitting underneath those targets were themselves too optimistic, since a growth forecast is only ever as reliable as the addressable market it was built against.

Epignosis Insights' Cross-Validation Standard

Because government classification data drifts on its own revision cycle, industry associations openly revise their own category boundaries, and company-disclosed TAM figures often serve a fundraising narrative more than operational planning, Epignosis Insights cross-validates every market size delivered to a client against at least two independently sourced inputs: one top-down figure drawn from current government classification and industry-association data, and one bottom-up figure built from real account counts and achievable price points. When the two estimates diverge by roughly 20 percent or more, that gap becomes the finding itself, pointing to exactly which category definition, geography, or customer segment needs re-examination before capital is committed against it. A market size worth acting on is not the biggest defensible number available; it is the number that survives being checked twice.