Market Size Estimate

You size a market the way an investor would want to see it: a precise definition, a bottom-up build from countable customers and real prices, a top-down cross-check, and a range that shows which assumption matters most. Every number has a source or is labelled as an assumption with its reasoning. A defensible range beats an impressive single figure that falls apart under one question.

When to use

• "How big is the market for our product?"
• "We need TAM, SAM, SOM for the pitch deck."
• "Is this market big enough to be worth entering?"
• "Size the opportunity in Germany versus the UK."
• Checking a market size figure someone else produced.

What you need

• The product, its price or pricing model, the target customer, and the geography.
• The purpose: an investor deck, an internal go or no-go, a budget. This sets how much precision is worth paying for.
• The owner's own data if any: customer counts, conversion rates, average deal size, sales capacity (from a connected CRM such as HubSpot or Salesforce, or a sheet).
• web_search, web_extract and your browser for official statistics, company filings and public pricing pages.

Method

1. Define the market precisely, in one sentence covering: who buys (customer type and size), what they buy (the product category, not the whole industry), where (geography), the unit (annual spend in a stated currency) and the year. "Annual spend by UK dental practices on patient-booking software, 2026, GBP" is a market. "Healthcare software" is not.
2. Set the three layers:
   • TAM (total addressable market): annual revenue if every customer who could use this category bought it
   • SAM (serviceable available market): the part of TAM your product and go-to-market can actually serve today (segments, geographies, languages, channels, integrations)
   • SOM (serviceable obtainable market): the share of SAM you can realistically win within a stated period, usually 3 to 5 years
3. Bottom-up first (it is the most defensible):
   • number of potential customers, from official statistics where possible: business counts by industry code and size band (for example the US Census Bureau's County Business Patterns, Eurostat structural business statistics, the UK ONS business population data), or population counts for consumer markets
   • times the share that has the need (sourced or a labelled assumption)
   • times the annual price per customer (your pricing, or competitors' public pricing pages, dated)
   • TAM = customers x share with the need x annual price
   • SAM = TAM restricted to the segments you serve now
   • SOM = SAM x achievable share, justified two ways: by capacity (sales people x deals per person per year x average deal value, or marketing funnel maths from your own conversion rates) and by comparison (shares that similar companies reached in a similar period, sourced)
4. Top-down as a cross-check: start from a published total (government data, an industry association, segment revenue in public company filings) and narrow it with sourced percentages. Treat headline figures from paid market-research press releases as weak evidence: their methods are usually hidden and estimates for the same market often differ by multiples. If you use one, quote it as "reported by" with the date, and never as the only basis.
5. Triangulate. Compare bottom-up and top-down. If they differ by more than about two times, find out why before presenting anything: usually a different market definition, geography or year. Explain the reconciliation.
6. Build the range. For each key input set low, base and high values with reasons. Show the result for each, and a sensitivity line showing which input moves the answer most. Do the arithmetic in a spreadsheet or with execute_code, not by hand.
7. Growth, if asked. Use CAGR = (end value / start value)^(1 / years) - 1 on sourced historical data, and name the drivers. Forecast growth is an assumption, and is labelled as one.
8. Write it up with every input in a table: value, source or assumption, date, grade of evidence.

Free sources by need

• Business counts by industry and size: national statistics offices and business registers; Eurostat for the EU
• Household and population counts: census data, national statistics offices, the World Bank and UN data portals
• Industry revenue: segment reporting in public company annual reports and filings (SEC EDGAR, Companies House, national registers)
• Prices: competitors' public pricing pages, with archived copies for history (Wayback Machine)
• Adoption rates: official technology-use surveys from statistics offices, peer-reviewed surveys, the owner's own funnel data
• Growth: time series from statistics offices; earnings calls of large players for direction, not numbers

Output

# Market size: <one-sentence market definition>
Date: <date>   Currency and year: <GBP, 2026>

## Answer
TAM <low to high, base>; SAM <...>; SOM <... by year N>. <One sentence on confidence and the input that matters most.>

## Bottom-up
| Input | Low | Base | High | Source or assumption |
|---|---|---|---|---|
| Potential customers | | | | <source, date> |
| Share with the need | | | | <source or "Assumption: ... because ..."> |
| Annual price | | | | <pricing page, date> |
| TAM | | | | calculated |

## Top-down cross-check
## Reconciliation
## SOM logic (capacity and comparison)
## Sensitivity
## Where the evidence is thin
## Sources

Put the TAM, SAM and SOM ranges on a show_card, and attach the spreadsheet.

Checks before you finish

• The market definition names customer, product, geography, unit, currency and year.
• Every input is either sourced (with date) or labelled as an assumption with reasoning.
• Bottom-up and top-down were both done, and any gap over about two times is explained.
• SOM is justified by capacity and by comparison, not just "we'll take 1%".
• The answer is a range with a base case, and the most sensitive input is named.
• Arithmetic was checked by formula, not mental maths.

Pitfalls

• The 1% fallacy. "If we get just 1% of a huge market" says nothing about how you would win it. Build SOM from capacity.
• Market defined too broadly. Using the whole industry's revenue inflates TAM by orders of magnitude and loses credibility instantly.
• Mixing years, currencies or geographies in one calculation.
• Quoting analyst headlines as fact. They are often unverifiable and contradictory. Cite carefully or not at all.
• False precision. "£2,347,912,000" implies certainty you do not have. Round to two significant figures and give the range.
• Sign-off. The owner decides how the figures are used. Figures going to investors or into financial plans should be reviewed by the owner or finance lead, who answers for them.

Worked example: references/worked-example.md.