Startup market size: build a bottom-up estimate investors can check

Define who pays, and what you count

Write one line: we sell this product to this buyer. Add the place and the need. Name who pays, not just who uses it.

A firm may have ten sites but one buying team. Count the firm once if it pays one fee. For a price per site, count sites instead. Do not mix the two.

Sequoia asks founders to name their customer and market. It also asks about other ways to solve the need. [3] Include firms that use a rival or do the work by hand. Do not assume they will switch.

• Buyer: who signs and pays?
• Unit: a firm, site, seat or sale?
• Scope: which needs and places fit?
• Exclusions: who cannot use this product yet?

Count buyers and test the price

Start with a local statistics office or trade body. Check what its count means. A list of sites is not a list of firms. Save the year, place and source link.

SBA separates market size from demand, location and price. It suggests both existing data and direct research with buyers. [1] A count can show scale. Buyer talks can test the need and price.

Use a yearly price for a yearly market estimate. State the currency. Keep small and large buyers in separate groups. Use their own prices. Label a planned price as a guess until tested.

• Record the buyer count and its definition.
• Remove the same buyer listed twice.
• Link the price to paid deals or price tests.
• Mark each input: fact, test or guess.

Show the maths, then narrow the scope

Here is an invented software example in US dollars. It is not real market data. Assume 2,000 firms and a $1,200 yearly fee.

2,000 × $1,200 = $2,400,000 in possible yearly spend. This is not a claim that all firms will buy.

Now apply clear limits. Remove firms your product cannot serve. Check place, language, needs and support. Assume those checks leave 300 firms. Record each cut and why it was made.

300 × $1,200 = $360,000 in possible yearly spend. This is the smaller scope you can serve. It is still not sales.

TAM means the full market within your stated scope. SAM is the part your product can serve. SOM is the part you aim to win over a set time. Explain the limits behind each. Do not claim an easy one per cent share.

Keep the sales plan apart from market size

A market count does not say how fast you can sell. Build a plan from contacts, meetings, paid tests and time to close. Show how many clients your team can support.

In the same invented case, a plan for 40 paid clients is unproven. At a full year each, 40 × $1,200 = $48,000. That is yearly spend, not first-year sales or cash received. Deals signed late will add less that year.

Test one weak input at a time. If the price is $900, the same 300 firms imply $270,000. Keep the buyer count fixed for that test. Then test fewer buyers on a separate line.

For a marketplace, total sales through the site are not its fee income. Show the share you keep. For seat-based fees, show seats per buyer. Do not force every model into one fee per firm.

Give investors a short proof sheet

Make a sheet with a row for each input. Add its source, date, scope and status. Link the raw count, each cut and the final sum. A reviewer should be able to repeat your maths.

Use a broad market report as a sense check. It may count different buyers, places or products. Explain the gap. Do not average two sums with different meanings.

AI can help find sources or spot repeat names. Open the source yourself. Check each key number and keep the link. Do not treat an AI answer as a data source.

For a global market, build each country or region on its own. Use local buyer counts and prices. State the exchange rate and date when you combine them. Do not scale one local guess to the world.

• Input and unit
• Source link and data year
• What is included and left out
• Fact, test or guess
• What new proof would change the input

Match the proof to your stage

At pre-seed, show the buyer group and gaps to test. At seed, use paid deals to test price and sales pace. At Series A, show results by buyer group and region. Keep facts apart from plans at each stage.

End with the next test, not the largest number. Name the input that could change your plan most. Assign a person and a date to check it.

This guide uses source pages read on 12 October 2026. The figures above are invented teaching examples. They are not forecasts or funding claims. This is general information, not investment, legal or tax advice. Rules differ by country. Seek qualified advice for your raise.