How to do it
Define the buyer precisely enough to count them
Not "small businesses" — "dental clinics in India with two to six chairs". The test is whether you could, in principle, produce a list. If you cannot describe the buyer well enough to count them, the number you produce will be a guess dressed as arithmetic.
Count them from a source you can name
Government registries, industry association membership, marketplace seller counts, LinkedIn company filters, app-store publisher counts. Write the source next to the number. A TAM with sources survives a conversation with an investor; a TAM without them ends one.
Use a price someone has actually paid
Take it from a competitor's pricing page, a freelancer's rate for the same outcome, or the salary cost of the person doing it manually today. A hoped-for price makes the entire calculation hypothetical.
Multiply, then write the assumptions underneath
The product of two numbers is less useful than the two numbers. List them separately so each can be challenged and corrected as you learn — the assumptions are the part that changes, and they are what makes the figure a model rather than a claim.
Worked example
Where to count buyers, by who you sell to
Accuracy falls as you move down the table, and so should your confidence. A TAM built on a professional register is worth arguing about; one built on consumer penetration assumptions is worth re-checking after the first ten conversations.
Where the numbers come from
Named sources you can open today. Advice to “check industry data” is what makes most articles on this subject useless.
- MCA / ROC company master dataFreeIndia
Every registered company in India by activity code, state and paid-up capital. The closest thing to a census of B2B buyers.
- GST taxpayer searchFreeIndia
Active GSTINs by state and business type — useful for counting registered traders and service businesses that the MCA misses.
- LinkedIn company filtersFree tier
Company count by industry, headcount band and geography. Run the filter, read the result count, ignore the list itself.
- Marketplace seller countsFree tier
Sellers on Amazon, Shopify app installs, WooCommerce usage from BuiltWith. The best free proxy for how many businesses of a type actually transact.
- App Store / Play publisher countsFree
Publishers in a category, for consumer and mobile-first markets where company registries tell you nothing.
- Industry association directoriesFree
Membership lists for regulated trades — clinics, pharmacies, brokers, agents. Small, accurate, and usually public.
- Government statistical portalsFree
MSME census, Economic Census, national statistics offices. Slow-moving but authoritative when a registry does not exist.
What an investor actually checks
Nobody reads the total first. An experienced investor looks at the two inputs and asks where each came from, because a large number built on two guesses is worth less than a small number built on a registry and a competitor's price list.
The second thing they check is whether the buyer definition matches the product you described five slides earlier. A TAM for "Indian SMBs" attached to a product that only works for clinics using one specific practice-management system is the most common inconsistency in a seed deck, and it undermines everything around it.
The third is whether you know what the number excludes. Being able to say "this excludes hospitals, because their procurement cycle makes them a different business" signals that the market was thought about rather than looked up.
When TAM is the wrong question
For a product sold to a small number of large buyers, TAM stops being informative. If there are forty possible customers in the world and each could pay a crore a year, the useful number is the forty, not the four hundred crore — the business is a sales problem, not a market-size problem.
Two-sided markets need two counts and a note on which side is the constraint. Counting only the demand side and multiplying by take rate produces a number that assumes supply appears for free, which is the assumption most marketplaces die on.
For usage-based pricing, annual revenue per buyer is a distribution rather than a figure. Use the median of a comparable product's public pricing tiers rather than the average, and say which you used — the mean is dragged upward by a handful of large accounts you will not win first.
What goes wrong
- Starting from an industry report — "the global healthcare IT market is $400B" tells you nothing about who would buy from you
- Counting everyone in a category rather than everyone with the problem
- Using the price you hope to charge rather than one someone already pays
- Presenting TAM as the opportunity — TAM is the ceiling, SOM is the opportunity
- Reporting one number with no assumptions, so nobody can tell you which input is wrong
Is your number plausible?
Each check catches a different way the arithmetic can be right and the answer still wrong.
- Divide TAM by the number of buyers. If the result is not a price a real customer would pay monthly, one of the two inputs is wrong
- Compare against the largest company already serving this market. If your TAM is smaller than their revenue, you have undercounted the buyers
- Ask whether you could produce a list of a hundred of these buyers this week. If not, the buyer definition is still too vague to count
Related answers
Should TAM be global or just my country?
Size the market you could actually sell into within about three years, and say which geography you used. A global TAM for a product with one language, one payment rail and one support timezone is not a ceiling you can approach, so it does not inform any decision you are about to make.
How often should I recalculate it?
After the first ten customer conversations, and then whenever an input changes materially — a new integration, a price change, a new geography. The total moving is less interesting than knowing which assumption moved it.
What if my category does not exist yet?
Count the buyers of the thing they use instead. A category with no incumbent still has people solving the problem somehow — with a spreadsheet, an agency, a freelancer or an in-house hire. Count those and price against what that alternative costs them.
Is a small TAM a reason to stop?
Only if it is smaller than the business you want. A ₹35 crore TAM cannot support a venture-scale outcome and can comfortably support a profitable team of six. The number is a decision input, not a verdict.