A TAM is not a slide with a big number on it. The figure that ends up in the board deck (total revenue if every possible buyer bought) is a sizing exercise, and it is useful for exactly one thing: deciding whether the market is worth entering.
The TAM that drives revenue is something else entirely: a living, enriched list of the specific accounts you could actually win, and it starts decaying the day you export it. Most teams confuse the two. They calculate a number, put it in a deck, and then go prospecting off a stale CSV that has nothing to do with it. This guide covers both: how to define and size a market (TAM, SAM, and SOM), and how to turn that number into an account list reps can work today.
What is total addressable market (TAM)?
Total addressable market is the total revenue opportunity available if every company that could possibly buy your product did. It is the outer boundary of demand, before you account for what you can realistically reach or win. TAM rarely travels alone. It comes with two tighter rings inside it, SAM and SOM, and the three only make sense in relation to each other.
The mistake is treating them as interchangeable. TAM tells an investor the ceiling. SAM tells a GTM leader the field they can actually compete in. SOM tells a sales team what to plan for this year. Confuse the rings and you either over-hire against a fantasy or under-resource against a real opportunity.
How to calculate TAM: top-down vs. bottoms-up
There are two ways to arrive at a TAM number, and they tend to disagree. The disagreement is the most useful output of the exercise.
Top-down starts from a published market figure and cuts it down with assumptions. Bottoms-up starts from a single account's value and multiplies up by the count of real accounts. When the two land far apart, one of your assumptions is wrong, and finding out which is worth more than either number on its own.
Enter your own inputs and watch top-down and bottoms-up TAM diverge
Top-down
Cut a published market figure down with assumptions
$3.36B
Top-down TAM = size × category × serviceable
Bottoms-up
Multiply one account's value by the count of real accounts
$340.2M
Bottoms-up TAM = accounts × ACV × penetration
Gap between the two methods
9.9x
One of these is fiction, almost always the top-down percentages
Top-down and bottoms-up are not two ways to get the same number, they are a cross-check. When they diverge by more than a few times, the top-down percentages are usually the fiction, and the bottoms-up account count is the number you can actually act on.
Bottoms-up wins for GTM because its inputs are things you can verify and then act on. A top-down model gives you $3.4B and no phone numbers. A bottoms-up model gives you 42,000 accounts that fit your ICP, and that count is the first draft of an actual list. The category-share percentages in a top-down model are where teams quietly insert optimism; the account count in a bottoms-up model is auditable, because you can go pull the accounts.
TAM, SAM, and SOM at a glance
The three rings, for a payroll tool serving US restaurants
| Term | What it measures | Example | What it's for |
|---|---|---|---|
| TAM (Total Addressable Market) | Total revenue if every possible buyer bought | All ~750,000 US restaurants x annual contract value | Sizing the opportunity; deciding whether to enter |
| SAM (Serviceable Addressable Market) | The slice your product and model can actually serve | The ~310,000 restaurants with 10+ employees on a supported POS | Setting GTM strategy and segment focus |
| SOM (Serviceable Obtainable Market) | The slice you can realistically win in a given period | The ~9,000 accounts your team can reach and close this year | Quota, headcount, and pipeline planning |
The numbers narrow fast, and that narrowing is the point. A TAM of 750,000 restaurants is a fundraising figure. The 9,000 accounts in your SOM are the ones a rep can name, call, and close. Everything useful happens as you move inward through the rings.
A TAM number is not a TAM list
The number sizes the opportunity; the list is the opportunity. A TAM of 42,000 accounts is a claim. The 42,000 rows, each with a domain, a headcount, a tech stack, a funding date, and a verified contact, are the thing a rep works. Most teams build the first and pretend they have the second.
The gap between them is enrichment. A market-size figure is one cell in a spreadsheet. A TAM list is every one of those accounts pulled from a real source, enriched with the firmographic and behavioral data that tells you which ones fit, scored against your ICP, and handed to sales with the context to act. The number takes an afternoon and an analyst. The list used to take months, which is why so few teams ever built one and instead worked off whatever export was handy.
That is the part that changed. Mistral AI mapped its global TAM (every qualified enterprise account across more than 100 countries) and turned the number into a worked list in a fraction of the time the old process took.
Qualified global enterprise accounts Mistral AI sourced in two weeks by mapping its TAM in Clay, down from a two-month process.
Read the full storyThe lesson is not that Mistral moved fast. It is that the deliverable was a list, not a number. The two-month process they replaced was the manual work of turning a market into accounts: finding the companies, enriching each one, deduplicating across vendors, and scoring for fit. That work is what separates a TAM you can cite from a TAM you can sell into.
Why a static TAM decays
A TAM list is accurate for about a day. The moment you export it, the market keeps moving and your file does not. Companies raise rounds, change headcount, get acquired, swap tools, and hire the exact role that signals they are ready to buy. A static export records none of it. Six months later you are prospecting off a snapshot of a market that no longer exists.
Decay is not a rounding error. It compounds, and it compounds in the fields that matter most for prioritization: funding recency, hiring signals, tech stack, and contact validity.
Play both timelines and watch a static TAM rot while a live one updates
Static export (built once)
0%
of accounts stale by month 0
Live TAM in Clay
100%
current — 0 of 3 promoted to T1 so far
A static TAM does not fail all at once. It decays fastest in exactly the fields you prioritize on (funding, headcount, contacts, stack), so within a year the majority of a once-accurate export is actively misleading. The same events, captured live, become timing instead of rot.
The fix is not to update the list more often by hand, which just moves the decay around. The fix is a TAM that re-enriches itself on a schedule and re-scores accounts when their signals change, so the list reflects the market today rather than the day you built it. A live TAM does not just stay accurate; it surfaces timing. When an account raises a round or posts the role your product serves, the same monitoring that keeps the record fresh also tells a rep this is the moment to reach out.
From a TAM map to prospecting
A live TAM is the input to prospecting, not a substitute for it. Once the market is mapped, enriched, scored, and kept current, the account list feeds directly into the motion that works it: the right accounts, in priority order, with the context a rep needs to open a conversation. The TAM answers which accounts exist and which fit. The complete guide to B2B prospecting is the motion that turns that ranked list into conversations and pipeline. The two are linked but distinct, and a clean handoff between them is what separates a market map that sits in a tab from one that produces revenue.
This handoff matters most when the market spans regions, because a TAM that looks complete in one country can be threadbare in another. Different markets have different data vendors, different coverage, and different rules, and a single global export papers over all of it. A TAM is only as real as its worst-covered segment. A live, multi-source TAM map lets you test coverage market by market and fill the gaps before reps ever touch the list, instead of discovering them mid-campaign.
Which accounts belong in the TAM at all is a separate, upstream decision: your ideal customer profile defines who counts as a fit, and the TAM is the full set of companies that match it. Sharpen the ICP and the TAM gets smaller and more workable; loosen it and you are back to sizing a market you cannot serve.
How to build a live TAM in Clay
Building a live TAM is a sequence, not a single export. You define the ICP that decides who belongs in the market, source the matching companies from the providers that actually cover your segment, enrich each account with the firmographic and behavioral data that drives scoring, score every account against your ICP, and then put the whole thing on a monitoring schedule so it re-enriches and re-scores itself as the market moves. The output is the list reps work, and it stays current without anyone rebuilding it.
Each of those steps has a specific build path in Clay, from picking the right data sources for your geography to writing the AI formulas that classify accounts into tiers. The full click-by-click workflow lives in the guide on how to build a TAM in Clay, which walks the entire sequence from ICP to a monitored, rep-ready market map. If your goal is narrower, two related playbooks go deep on the prospecting end: how to generate B2B leads covers turning a market into qualified pipeline, and how to build a targeted prospect list covers building a scored account list from a precise ICP.
If you are deciding where to start, start with the bottoms-up count. It is the number you can verify, it doubles as the first draft of your account list, and it forces the ICP precision that everything downstream inherits. A defensible account count beats an impressive market-size slide every time.