The TAM slide is almost always wrong.
Top-down market sizing is how decks justify ambition. Bottom-up is how investors check the math. The gap between the two is usually where the real conversation happens.
01The pattern
Every deck we receive contains a TAM slide. The number is typically large, round, and presented without provenance — “the global market for X is $200 billion by 2030.” Most investors learn early to discount these figures. Fewer learn to replace them with a defensible bottom-up rebuild that becomes the basis for an honest conversation about pricing power, market share, and the realistic upside of the business.1
Across 1,142 decks analysed in private beta, the median TAM claim reconciles to roughly 14% of a defensible bottom-up rebuild. That is not a rounding error. That is an order of magnitude.2
02Why decks inflate the number
Three structural pressures explain almost all of it:
- Pattern matching. Founders read the same blog posts as us. The $200B figure exists because the previous deck in the same category said $200B. The number has been laundered through enough decks that it now feels canonical.
- The “what investors want to hear” trap. Founders are correctly diagnosing that a small TAM kills the round. They are incorrectly concluding that a large unsourced TAM saves it.
- Top-down arithmetic is cheap. A bottom-up rebuild takes a week. A top-down claim takes an afternoon. The slide gets built in the medium that the time budget allows.3
03The four-cut rebuild
A defensible market size answers four questions in sequence. Each cut narrows the number. Each cut also makes it more useful — because it is now a real budget the business can compete for.
- Universe. How many entities of the right shape exist? (Firms, households, transactions — whichever unit actually buys.)
- Reachable. Of those, how many can plausibly buy this product, given regulation, language, integration prerequisites, or geography?
- Willing. Of those, what fraction has the unsolved pain at sufficient severity to actually move budget for it?
- Wallet. What do they currently spend on the closest substitute — and what's a credible capture rate against that wallet, not against revenue at large?
The output is usually one to two orders of magnitude smaller than the top-down number. That is the point. The new number is one we can defend at the IC table.
04A worked example
An ostensibly mid-market workflow tool claims a $120B TAM (“global workflow software, projected 2030”). A four-cut rebuild on the same business, using the founder's own ICP definition:
TAM reconciliation — claimed vs. rebuilt
N = 12,400 firms · ACV $84K · capture 8%The rebuilt number — $1.04B — is not embarrassing. It is the right number for the question the IC actually asks: “if this team executes well over five years, what does the business look like?” A $1B revenue-eligible market with 8% capture and an $84K ACV is a Series A worth doing. A $120B TAM was always going to fail the first follow-up question.4
05What makes a TAM defensible
Three properties separate a TAM number that survives the IC table from one that doesn't:
- It cites units, not dollars. “12,400 mid-market firms in North America with 50–500 employees and the relevant compliance posture” is a number you can argue with. “$120B by 2030” is not.
- It's tied to a substitute spend, not a category abstraction. The right denominator is “what these specific customers currently spend on the closest substitute,” not “the global software market.”5
- It survives a capture-rate sensitivity. A founder who can answer “what does this look like at 4%, 8%, 15% capture?” has a number. A founder who can only quote one figure has a slide.
06Field notes
A few patterns we see often enough to mention:
- Decks that open with a four-cut TAM are correlated, in our corpus, with materially higher follow-on signals from partners. The act of doing the work is itself diligence-signal.
- Founders are sometimes more uncomfortable with a $1B defensible TAM than with a $120B undefensible one, because they have been coached that “bigger is better.” This is almost always wrong at IC.
- When we run the four-cut rebuild in a dossier, we publish both numbers and the gap. The gap is information. Hiding it would be worse than the original.
07Methods
Citations
- Early Capital desk note, “Top-down vs. bottom-up sizing — failure modes,” internal memo, Mar 2026.
- EC corpus statistics, N = 1,142 dossiers, Jan–May 2026. Median computed on log-scale ratio of claim ÷ rebuild.
- Founder time-allocation survey, n = 84 founders self-reporting hours spent on TAM slide construction.
- EC dossier sample: 14 Series A rounds closed in private beta with rebuilt TAMs < $5B disclosed in IC memo.
- See also: Damodaran, A., “The Dark Side of Valuation: Top-Down vs. Bottom-Up,” Stern NYU working paper, 2024.
- Reviewed by EC IC board, May 24, 2026. Pre-publication review covered methodology, anonymisation, and sample representativeness.
Want the rebuild on your TAM slide?
The four-cut framework — and every other pattern in the research archive — is baked into every Early Capital dossier. Upload a deck and you get the reconciliation back in roughly fifteen minutes: claimed number, rebuilt number, the gap, and where it came from.
Kill flags: the eleven we see most often.
A taxonomy built from 1,142 dossiers. Eleven flags account for 78% of all material kills — and four are recoverable if the founder addresses them in the first call.
Read note →The cohort table the deck didn't include.
Why founders show triangle charts instead of cohort tables — and the three reconstructions that turn a flattering retention slide into a defensible one.
Read note →Three questions every angel should ask, but rarely does.
Most angel diligence calls run forty-five minutes and never get past traction. These three questions surface more than the other forty-five combined.
Read note →