Market sizing is the exercise of estimating how many companies could buy your product and what that demand is worth in revenue. Founders size markets for pitch decks, sales leaders size them to carve territories and set quotas, and marketers size them to decide which segments deserve budget.
The math involved is multiplication. The work is in the counting. A market size is only as good as the list of companies underneath it, and that's a data problem before it's a strategy problem.
What do TAM, SAM, and SOM mean?
The standard output of a market sizing exercise is three nested numbers, each a subset of the one before it.
TAM (total addressable market) is everyone who could conceivably buy the product category. If you sell field service software, TAM is each company that dispatches technicians, anywhere you could ever operate.
SAM (serviceable addressable market) is the slice your product and business model can serve today. Filter TAM by geography, company size, language, vertical, and anything else that makes an account a real prospect rather than a theoretical one.
SOM (serviceable obtainable market) is what you can win within a planning horizon, given competitors, switching costs, and how many deals your team can close. It's the number that should connect to next year's revenue plan.
Investors read three. A giant TAM with a hand-waved SOM says the founder hasn't done the work. A modest TAM with a credible, list-backed SOM says the opposite.
Top-down or bottom-up: which method holds up?
Verum provides a written, scope-specific charge and schedule after reviewing the records, requested checks, exception rules, and delivery format. Analysts publish those headline figures for whole categories, which rarely line up with what one product does.
Bottom-up sizing counts actual companies. Define the profile that buys from you, count how many exist, multiply by what an account pays per year. Three inputs, each checkable:
- The profile. Which industries, which size bands, which geographies. This is your ideal customer profile written as filterable criteria.
- The count. How many companies match. This comes from data, and it's where most models quietly go wrong.
- The charge. Annual agreement value, ideally from your own closed deals rather than aspiration.
Bottom-up wins because the artifact behind it is an account list. When the board asks where growth comes from, you can open the file and point at names. A top-down model can't answer that question.
Where do the company counts come from?
For US markets, County Business Patterns is a useful starting point because it publishes establishment counts by industry and geography. Record the release year, geographic unit, and industry-code version used in the model; a company count assembled from a different unit or vintage is not directly comparable.
Census establishment counts do not map directly to buying decisions: a multi-location organization can appear as several establishments while purchasing remains centralized. Commercial sources add filterable firmographic fields, but coverage and classification methods differ. Compare each source against the same dated sample before using its count in a market model.
Serious sizing work triangulates: government counts for the ceiling, commercial data for the filterable list, and spot verification to measure how much of the list is real.
Why does bad firmographic data wreck the estimate?
Each filter in a bottom-up model runs on a data field, and each field has an error rate. Industry codes get self-reported or vendor-guessed. Employee counts go stale as companies grow, shrink, or get acquired. Revenue for private companies is modeled, and models miss. Locations lag each office move.
Worked example: suppose industry, size band, and geography each contain some classification errors. Combining the filters compounds the uncertainty because a company can fall on the wrong side of any one boundary. Report sampled error by field and show how exclusions change the counted market.
Two teams can size the same stated market differently when their underlying company lists, eligibility rules, locations, hierarchy treatment, or source dates disagree. Reconcile those inputs before comparing the totals.
Example
Worked example: start with a vendor file of US banks filtered by asset band, then reconcile the institutions against current FDIC records. Remove closed or merged institutions, separate credit unions from banks, and calculate the market from the reviewed count. The useful output is not a dramatic estimate; it is a list whose inclusion rules, source date, exclusions, and unresolved records can be examined during diligence.
Common Mistakes
- Sizing the category instead of the product. Quoting the analyst figure for "the data market" when you sell one workflow to one vertical. Count your buyers, charge your product.
- Counting establishments as accounts. Locations, subsidiaries, and franchises inflate counts. Deduplicate to the buying entity before multiplying by ACV.
- Filtering unverified fields. A size-band filter on stale headcounts silently reshapes the market. Verify fill rate and correctness on each field the model filters by.
- Leaving the model stale. Re-run it when the source releases a new vintage or when your segment definition changes. Keep prior inputs so changes in the market can be separated from changes in methodology.
Frequently Asked Questions
What is market sizing?
Market sizing is estimating how many companies could buy your product and what that demand is worth in revenue. B2B teams usually express the result as three nested numbers: TAM (total addressable market), SAM (serviceable addressable market), and SOM (serviceable obtainable market).
What do TAM, SAM, and SOM stand for?
TAM is the total addressable market: everyone who could conceivably buy the product category. SAM is the serviceable addressable market: the slice your product and business model can serve today, filtered by geography, segment, and fit. SOM is the serviceable obtainable market: the share of SAM you can win in a planning horizon given competition and your capacity to sell.
What is the difference between top-down and bottom-up market sizing?
Top-down starts from a published industry figure and narrows it with assumptions. Bottom-up counts companies that match a defined profile and applies an observed account-value range. Preserve the account list, source date, eligibility rules, and assumptions so the result can be reviewed.
How do you calculate TAM for a B2B company?
Define who buys, count them, and multiply by annual agreement value. The count usually starts from industry codes: pull establishment counts from Census data for your NAICS codes, or build the list from commercial databases, then multiply by what an account is worth per year. The multiplication is trivial. Getting an accurate count of companies that match your profile is the hard part.
Why does data quality change a market size estimate?
Because each filter in a bottom-up model runs on firmographic fields: industry, employee count, revenue, location. Measure change against a dated baseline instead of applying a general decay benchmark, then set the review cadence from the changes observed in your own system. The error compounds across filters, and the final number can be off by a third without looking wrong.
Can Verum build the dataset behind a market sizing model?
Yes. You get a countable, filterable list instead of an estimate, and the same file works as a targeting list afterward.
Related Terms
Related: All Glossary Terms | TAM SAM SOM Analysis | Enrichment Services | Cleaning Services