Verified against ChatGPT · 2026-08-12
Size a market without letting ChatGPT quietly invent the number in the middle of the math
Builds a market-sizing analysis where every multiplier and assumption is stated and sourced separately, so a confident-looking TAM figure doesn't hide an invented number two steps upstream.
The prompt
Ready to copy — highlighted parts are example details you can swap.
Help me size this market, but every number in the final figure has to trace back to either a cited source or an explicitly labeled assumption — no unlabeled numbers just appearing in the middle of the math. MARKET TO SIZE The addressable market for a B2B SaaS tool that automates expense report categorization for companies with 50-500 employees in the US. NUMBERS I ALREADY HAVE US Census data shows roughly 45,000 companies in the 50-500 employee range; our current customer data shows average contract value of $8,400/year. GAPS I EXPECT YOU'LL HAVE TO ESTIMATE No public data on what percentage of those companies currently use any automated expense tool versus manual processes. APPROACH PREFERENCE Bottom-up, starting from the number of eligible companies and our known average contract value. Build the sizing using Bottom-up, starting from the number of eligible companies and our known average contract value. (top-down from a broader known market, or bottom-up from unit economics — whichever I specified). At every step, label each number as either "sourced" (with the source named) or "assumed" (with the reasoning for the assumed value stated plainly, not just the number). If you have to estimate something not covered in US Census data shows roughly 45,000 companies in the 50-500 employee range; our current customer data shows average contract value of $8,400/year. or No public data on what percentage of those companies currently use any automated expense tool versus manual processes., flag it explicitly as a new assumption rather than quietly inserting a plausible-sounding figure. Where an assumption meaningfully drives the final number, note how sensitive the final TAM is to that one assumption — if changing it by half would roughly halve or double the output, that needs to be visible, not buried in a single pass-through calculation. WHAT NOT TO DO Do not present a single polished TAM figure with the underlying assumptions relegated to a footnote no one will read — the assumptions are the most important part of this output, not an afterthought. Do not round or smooth an estimate to make it look more precise than the underlying data supports. OUTPUT FORMAT 1. The sizing calculation, step by step, each line labeled sourced or assumed. 2. A final TAM range (not a single false-precision number) with the low and high bound explained by which assumptions drove each end. 3. The single assumption the final number is most sensitive to.
Customize
Optional — swap in your own details for the highlighted parts above.
Why this works
Market-sizing exercises are especially prone to a specific failure mode where a model chains several multiplications together and, when it hits a step with no clean data, fills the gap with a plausible round number — a "15% adoption rate" or "roughly 20% of companies" — that reads as confidently as the sourced figures around it, because the output format doesn't distinguish a cited statistic from a made-up placeholder once they're both just numbers in the same equation. Requiring every line to be explicitly labeled sourced or assumed breaks that camouflage: an assumed number sitting next to its stated reasoning invites scrutiny in a way the same number embedded silently in a multiplication chain never would. The sensitivity-flagging requirement matters because market sizing math compounds multiplicatively, so a single soft assumption near the start of the chain can swing the final TAM by multiples, and a single polished output number gives no signal about which input to actually go verify if the number needs to hold up under real scrutiny — asking specifically which assumption the output is most sensitive to turns a black-box calculation into something a reader can prioritize checking. Requiring a range rather than a single figure directly counters the false-precision problem: a bottom-up calculation built partly from assumptions cannot honestly produce a number precise to the dollar, and presenting one anyway implies a level of certainty the underlying inputs don't support.
What you get back
45,000 eligible companies [sourced: Census data] x $8,400 average contract value [sourced: internal data] x est. 30% realistic penetration ceiling within 5 years [assumed: comparable SaaS categories typically plateau between 20-40% penetration] = roughly $113M-$151M TAM range. Most sensitive assumption: the 30% penetration ceiling — at 20% the TAM drops to roughly $76M, at 40% it rises to roughly $151M.
Verified against
ChatGPT GPT-5.1 · 2026-08-12
Changelog
- 2026-08-12 — Initial publish, verified against ChatGPT GPT-5.1.
Need this built into your business?
If a prompt isn't enough — what Scult builds, built and maintained for you — that's Scult's day job.
EXPLORE WHAT SCULT BUILDS
