Market Sizing That Survives Diligence

Jagdeep Singh · Jagdeep Ventures · Last revised July 2026

Most market sizes in pitch decks share a genealogy: an analyst report headline, a percentage plucked for "serviceable," and a footnote nobody can reconstruct. The number survives until the first person with an incentive to check it, a diligence analyst, an investment committee, a corporate development team, checks it. This note describes the construction we use so the number survives that meeting.

The failure mode

Top-down sizing fails not because aggregates are wrong but because the filters applied to them are unstated and unverifiable. "The market is $40B and we need 1%" contains no information about demand for your product. Its arithmetic is unfalsifiable, which is precisely why sophisticated audiences discount it on sight.

Triangulate: three independent constructions

Top-down, done honestly. Start from an authoritative aggregate, then decompose with explicit, sourced filters: geography, segment, use case, price tier. Every filter is a stated assumption someone can challenge, which is the point. The chain of reasoning is the deliverable, not the terminal number.

Bottom-up, from the unit of demand. Count the buyers (registries, census data, platform counts), estimate adoption and frequency, multiply by realistic pricing. A Fermi decomposition forces each factor into the open. This is the estimate a diligence team will attempt to rebuild; if you have already built it, the meeting goes differently.

Demand-side, from displaced spend. Whatever you sell replaces money or time currently spent somewhere: budget lines, headcount hours, incumbent tools. Sizing the displacement anchors the estimate in observed behavior rather than projected enthusiasm, and it catches the classic error of sizing a "market" nobody currently pays anything into.

The reconciliation is the insight

The three estimates will disagree, usually by 2–5×. Amateurs average them; the useful move is to interrogate the disagreement, because it is always caused by an identifiable assumption: a filter too generous, an adoption rate imported from a friendlier analog, a price the demand-side data will not support. Resolving the disagreement is where the actual learning about the market happens. In our experience the reconciliation memo changes strategy more often than the final number does.

Report a range, with drivers

The deliverable is a range with a stated confidence level and the two or three assumptions that dominate the spread, made visible with simple Monte Carlo sampling over the uncertain parameters, or even a low/base/high table built by hand. "SAM is $220–410M, driven mainly by enterprise adoption rate and realized price" invites a productive argument about adoption and price. "$1.3B" invites either belief or dismissal, neither of which improves the decision.

Red flags a reviewer will catch (so catch them first)

• Percent-of-a-big-number logic anywhere in the chain ("just 1% of...").
• A CAGR imported from a vendor report whose methodology you have not read.
• SOM defined by ambition ("what we plan to capture") rather than by capacity and go-to-market arithmetic.
• Adoption benchmarks from analogies with structurally different switching costs or purchase processes.
• A market that requires simultaneous behavior change from two sides of an ecosystem, sized as if one side decides alone.

None of this makes a market size true, estimates about the future never are. It makes the reasoning inspectable, which is the property diligence actually tests for. Numbers with visible reasoning get argued with; numbers without it get discarded. In fundraising and in strategy, being argued with is the win condition.