Revenue Builds: Top-Down TAM vs Bottom-Up Units
A revenue forecast built by shrinking a huge market size looks confident but hides its assumptions, while one built up from units, prices and capacity is slower to construct but far easier to sanity-check.
Ask an analyst how big a new company's revenue could get, and there are two honest ways to answer, and one lazy way that looks like the first honest way. "The global market for this product is $50 billion, and we'll capture 2% of it" sounds rigorous, but it's really just a guess about market share dressed up in a big scary number. The more defensible route builds the forecast from the ground up: how many salespeople, how many customers each one can close, how much each customer pays.
These are the top-down and bottom-up approaches to a revenue build, and knowing which one a model actually uses — regardless of which one it claims to use — is one of the fastest ways to judge whether a forecast deserves any trust.
Top-down starts from a total addressable market and shrinks it by an assumed share. Bottom-up starts from the smallest unit of the business — a salesperson, a store, a customer — and multiplies up. Bottom-up is more work, but every assumption in it is checkable against something real; a top-down share assumption usually isn't.
The two builds
Top-down (TAM-based): Revenue = Total Addressable Market × Assumed Market Share. This is fast and useful for a first sanity check on order of magnitude, but the entire forecast hinges on one number — the share assumption — that is usually asserted rather than derived from anything operational.
Bottom-up (unit-based): Revenue = Number of Units (salespeople, stores, customers, subscribers) × Output per Unit (deals closed, sales per store, revenue per customer) × Price. Each input can be checked against the company's actual current operations, hiring plans, or store rollout schedule, which makes the forecast auditable piece by piece.
Worked example
Top-down: A pitch claims a $40 billion TAM for a new payments product and forecasts 1% share within five years, implying $400 million of revenue. There is no way to independently check whether 1% is realistic without more information — it's a plug.
Bottom-up, same company: The sales plan calls for 80 sales reps by year five, each closing 15 new merchant accounts a year, each merchant processing $2 million of annual volume at a 15 basis point take rate. Revenue = 80 reps × 15 merchants × $2,000,000 × 0.0015 = $3.6 million per rep-cohort-year, aggregated across ramping headcount to roughly $290 million once you account for reps hired mid-year still ramping. This number can be checked against actual hiring plans, historical rep productivity, and current merchant economics — and it happens to land meaningfully below the top-down guess, which should make an analyst distrust the $400 million pitch.
What this means in practice
Analysts building or reviewing a DCF should always ask "which build produced this revenue line?" A number that only survives as a TAM-times-share calculation is a red flag for a pitch deck; a number built bottom-up from reps, stores, or subscriber cohorts can be stress-tested against reality — can the company actually hire that many reps, does the market have that many merchants left to sign.
The two methods often get run in parallel purely as a sanity check, and that's legitimate — the mistake is presenting a top-down TAM calculation as if it were evidence, rather than as the loose upper bound it actually is.
Further reading
- Damodaran, Narrative and Numbers (ch. 4, revenue growth)