There are two reasons to calculate a total addressable market, and they need different numbers. One is to tell a funding story, where a large figure supported by an analyst report does the job. The other is to decide how many reps to hire, which territories to draw and which segment to attack first — and for that, a number built from a research firm’s market estimate is useless.
This guide builds the second kind. It counts real accounts that match your profile, prices them at what you actually charge, strips out what you genuinely cannot serve, and then reconciles the result against the capacity of the team you have. That last step is the one almost every total addressable market article leaves out, and it is where the number finally starts changing decisions.
TAM, SAM and SOM, without the fog
Three numbers, three different jobs. Most confusion comes from teams quoting one and acting on another.
| Measure | Definition | Formula | Decision it drives |
|---|---|---|---|
| TAM | Every company that could ever buy, at 100% share | Matching accounts × average contract value | Is the category worth building in? |
| SAM | The part you can legally, technically and commercially serve today | TAM minus unreachable segments | Where do we sell this year? |
| SOM | What your current team can realistically capture | SAM × achievable share, capped by capacity | Headcount, quota, territory design |
The trap is treating these as three slices of the same pie chart. They are not. TAM and SAM are properties of the market, while SOM is a property of your team, and it changes the moment you hire or lose a rep.
Why top-down total addressable market figures fail operationally
Top-down sizing starts from a published market figure and multiplies it by a share you believe you could win. It is fast, it produces a large number, and it collapses the moment anyone asks which companies are in it.
- The analyst’s category boundary is not your product boundary, so the figure includes spend you can never win
- It gives revenue but no account count, and account count is what territories are made of
- It cannot be segmented by the attributes in your ideal customer profile, so it cannot tell you where to start
- It ages badly, and there is no way to refresh it yourself
Bottom-up sizing inverts this. You count the companies that match your filters, multiply by what those companies pay, and end up with a total addressable market you can drill into, sort, and hand to a rep as a list. The headline figure is usually smaller and always more defensible. Keep the top-down number for the board deck if you must, but never plan against it. HubSpot’s primer on TAM, SAM and SOM is a reasonable refresher on the definitions themselves.
The bottom-up method, step by step
Six steps, and the whole thing takes an afternoon once the profile exists.
- Write the filters. Industry codes, headcount bands, geography, and any technographic requirement. If you cannot express the profile as filters, size nothing until you can.
- Count accounts by segment, not in total. Run the same query at each headcount band separately, because pricing differs by band and a single blended number hides everything useful.
- Price each segment at realised ACV. Use the median of closed-won deals in that band over the last year, net of discount. List price inflates the total addressable market by 20–40% at most companies.
- Audit fifty accounts by hand. Sample the returned list, check each one really matches, and record the error rate. A 12% false positive rate means the count is 12% high.
- Apply the haircut. Reduce the count by the measured error rate rather than by a comfortable guess.
- Multiply and sum. Segment counts times segment ACV, added together.
Step four is the one people skip and the one that determines whether anyone trusts the result six months later. Every database returns some companies that closed, merged, or were classified under the wrong industry code, and the way to handle that is to measure it rather than to argue about it. Our guide to B2B data accuracy covers how to run the audit properly.
A worked total addressable market calculation
Take a compliance software company selling to B2B services firms in the US, UK and Germany. The filters return the following, after a 9% haircut from the fifty-account audit.
| Segment | Matching accounts | Median realised ACV | Segment value |
|---|---|---|---|
| 50–199 employees | 18,400 | $12,000 | $220.8M |
| 200–999 employees | 6,200 | $45,000 | $279.0M |
| 1,000+ employees | 1,150 | $120,000 | $138.0M |
| Total addressable market | 25,750 | — | $637.8M |
Two things are immediately visible that a single blended figure would have hidden. The mid-market band carries the most value despite holding a quarter of the accounts, and the enterprise band is small enough that it is a named-account motion rather than a territory. That is a segmentation decision falling straight out of the arithmetic.
From TAM to SAM: what actually removes accounts
SAM is not TAM multiplied by a confidence percentage. It is TAM minus specific, nameable exclusions, each of which you should be able to point at in the data.
| Exclusion | Typical impact | Reversible? |
|---|---|---|
| Regions with no lawful basis or no local-language support | Removes 20–35% of accounts | Yes, with investment |
| Incompatible tech stack or no required integration | 5–15% | Yes, on the product roadmap |
| Below the price floor once discounting is honest | 5–20% of the smallest band | Only with a lower-touch motion |
| Existing customers | 1–5% | No, but they become expansion pipeline |
| Locked into a competitor contract | Timing constraint, not a permanent one | Yes, at renewal |
Apply those to the example and the reachable set falls to 11,800 small, 4,200 mid-market and 700 enterprise accounts, a SAM of roughly $415M. The value of writing exclusions down individually is that each one has an owner: language coverage belongs to marketing, integrations to product, the price floor to finance. A SAM built this way doubles as a roadmap of what would grow it.
SOM is a capacity calculation, not an ambition
Here is the section that changes plans. Multiplying SAM by 5% because that feels achievable tells you nothing. Work forward from what the team can physically touch.
- 8 reps, each working about 150 accounts per quarter across all channels
- 4,800 account-quarters a year, but an account tolerates roughly two sequences annually, so about 2,600 distinct accounts get touched
- At a 5% account-to-meeting rate, that is 130 first meetings
- At 25% meeting-to-close, about 33 new customers
- At a $50,000 blended ACV in the bands reps actually prioritise, roughly $1.65M in new ARR
That is a SOM of about 0.4% of SAM, and it exposes the real constraint: the team can only touch 2,600 of 16,700 reachable accounts a year, so 84% of the serviceable market never hears from anyone. The lever that matters is not a bigger total addressable market, it is coverage — which accounts get the 2,600 slots. That is exactly what territory planning and prioritisation decide, and why lead scoring earns its keep long before you need more headcount.
Run the same arithmetic backwards to size hiring. If next year’s target is $5M new ARR at that conversion profile, you need roughly 100 wins, 400 meetings and 8,000 distinct accounts touched — which needs both more reps and a SAM comfortably above 20,000 accounts. If your SAM cannot supply that, the constraint is the market, and no hiring plan solves it.
Where the account counts come from
Three sources, used together, because each is wrong in a different direction.
| Source | Strength | Known bias |
|---|---|---|
| Prospecting database counts | Matches your exact filters, contactable records | Under-counts small and non-digital firms |
| Government statistics | Complete population by industry and size band | No contacts, coarse categories, published in arrears |
| Job postings and hiring data | Proves the function you sell to exists | Biased towards companies currently growing |
For the population check, the US Census Bureau’s Statistics of U.S. Businesses gives firm counts by industry and employment size class, and the NAICS code system tells you which categories you are actually asking for. If a database returns 18,400 accounts where official statistics suggest 40,000 firms exist in that band, the gap is coverage, not market size, and your total addressable market should reflect the larger number while your list is built from the smaller one.
Mistakes that inflate a total addressable market
- Counting employees instead of companies. Per-seat pricing tempts teams to multiply headcount by list price, which assumes universal adoption inside every account.
- Sizing globally while selling locally. If support runs one timezone and the contract is in one language, the rest of the world is TAM, not SAM.
- Double counting corporate structures. Parent, subsidiary and trading names appear as three records and one buyer. Deduplicate on domain before counting.
- Using list price. Realised ACV after discount is usually materially lower, and it is the only number that has ever been paid.
- Treating the total addressable market as pipeline. It is the ceiling of demand, not a forecast, and the distance between them is capacity.
When to recalculate
Annually is the default answer and it is wrong, because the inputs move on their own schedules. Recalculate on triggers instead.
| Trigger | What it changes | Scope of rework |
|---|---|---|
| Pricing or packaging change | ACV per segment | Re-price, keep the counts |
| New region or language | Exclusions, so SAM only | Re-run filters for that geography |
| New product or module | Both counts and ACV | Full rebuild |
| Profile revision after a win-rate review | Which accounts qualify | Re-run filters, re-audit fifty |
| Quarterly data refresh | Count drift of a few percent | Note it, do not rebuild |
One more source of growth is worth planning for deliberately: when the profile itself proves too narrow, the fastest legitimate expansion is usually adjacent rather than larger. Finding companies that resemble your best customers on dimensions your filters never captured, through lookalike company search, tends to add accounts with a higher conversion rate than simply widening a headcount band.
Frequently asked questions about total addressable market sizing
Is a total addressable market measured in revenue or accounts?
Both, and the account count is the one operators need. Revenue answers whether the category is big enough to justify the company; the number of matching accounts answers whether there is enough work for the sales team, how many territories you can draw, and how quickly the list will be exhausted.
What counts as a good number?
It depends entirely on the funding model. Venture investors generally want a credible path to a billion-dollar category, whereas a bootstrapped business can build a strong company inside a $50M market it dominates. The more useful test is the ratio: if your current ARR is already several percent of SAM, growth has to come from expansion, new segments or new geographies.
How do I size a market that does not exist yet?
Size the budget you displace. Count the companies that currently pay for the manual process, the agency, or the incumbent tool your product replaces, and price at a defensible fraction of what they already spend. It is a hypothesis either way, but one anchored to observable spending rather than to a growth-rate projection.
Should the calculation include accounts we cannot find contacts for?
Include them in TAM and exclude them from the working list. An account with no reachable buyer is still part of the market, it is simply not part of this quarter’s coverage. Track the gap as a data problem with a percentage attached, because it is usually fixable.
Size the market you can actually work
A total addressable market that exists only as a figure on slide four settles arguments about ambition. One built from counted accounts, realised prices and honest exclusions settles arguments about hiring, territories and which segment gets next quarter. The second kind takes an afternoon longer to produce and is the only one that will still be referenced in six months, particularly once it is feeding the sales cadence your reps run every day.
ZenBee gives you the account counts behind the calculation in real time — filter or describe your market across 35M+ companies, see exactly how many accounts match at each size band, and turn the result into a working list rather than an estimate. Request a demo or see pricing.