There are two completely different activities that both get called TAM. One is the slide in a fundraising deck with a large number on it. The other is TAM mapping: the operational work of turning “our market” into a list of named accounts you can actually contact this quarter. The first is an argument. The second is infrastructure.
This covers the second. How to build the map, what the count tells you before you write anything, and why the number in the deck is almost never the number you should be working from.
Why the Deck Number Is Useless Operationally
One TAM is an argument. The other is infrastructure. Only one of them tells you how many domains to provision.
A top-down TAM — market size times a plausible share — is unfalsifiable and unactionable. It cannot tell you whether to run outbound or account-based selling, how many sending domains to provision, or whether your segment is large enough to survive a 30% annual data decay rate.
Bottom-up TAM mapping answers all three, because it is built from records rather than assumptions. The TAM, SAM and SOM distinction covers the framing; what follows is the build.
TAM Mapping Step 1: Turn the ICP Into a Filter
raw universe 2,410,000 ├ firmographic box 612,400 headcount, geo, funding stage ├ technographic match 188,900 running the tool you integrate with ├ hiring signal (90d) 41,700 roles that imply your problem └ deliverable contacts 38,220 ← the actual TAM
The input to TAM mapping is an executable filter, not a description. “Mid-market B2B SaaS” cannot be run against a database. This can:
- Headcount 50–500
- Raised a Series A or later in the last 36 months
- Running a specific category of tooling
- Has at least one person holding a named role you sell to
Data providers such as Cognism and Apollo will run a filter like this and return a count, which is all you need for the next step. Each condition should be there because it predicts something — win rate, deal size, retention — and you should be able to say which. Conditions added because they feel right are how a map narrows to a segment nobody validated. Negative ICP is the other half: the explicit exclusions usually remove more waste than any inclusion adds.
Step 2: Run It and Read the Count
| Accounts returned | What it means | Right motion |
|---|---|---|
| Under 200 | Not a market, a named list | Account-based, human research per account |
| 200–1,000 | Narrow but workable | Hybrid: tiered ABM plus light outbound |
| 1,000–20,000 | The outbound sweet spot | Systematic outbound with signal layering |
| 20,000–100,000 | Likely too loose | Re-segment before building |
| Over 100,000 | The filter is not a filter | Go back to step one |
This table is the single most useful output of TAM mapping, and it costs an afternoon. Teams routinely spend a quarter building outbound for a 300-account market that was never going to support it, and the count would have said so on day one.
Step 3: Segment Before You Enrich
Do not enrich the whole map. Enrichment costs scale linearly with records and most of the map is not where you should start. Split it into tiers by expected value — where you win fastest, retain longest and expand most — and enrich tier one only.
This also protects the sending infrastructure. Launching into your best-fit segment generates the engagement that builds domain reputation, so by the time you reach the looser tiers your domains have history. Launching into the whole map at once does the opposite.
Step 4: Layer Signal Onto the Map
A static map tells you who. It does not tell you when, and timing accounts for more of the variance in reply rate than almost anything in the copy. Funding events, executive hires, technology changes and hiring surges all mark the moment an account moves from theoretically qualified to actually reachable.
Operationally this means the map is a live query, not a spreadsheet export. Accounts should enter the sequence in the week they qualify. Intent data belongs here too — as a prioritisation layer on a validated map, never as a substitute for building one.
Step 5: Plan for Decay
Every serious data vendor publishes a decay figure and they cluster in the same place — HubSpot’s sales research among them. B2B contact data decays at roughly 25–30% a year, driven mostly by job changes, and a TAM map is worthless the moment you stop maintaining it. Budget a quarterly refresh of tier one and an annual rebuild of the whole map from the filter — which is only cheap if step one produced a filter you can re-run rather than a list someone assembled by hand.
Our note on list and data decay covers the refresh mechanics, and account intelligence versus intent data covers which signals are worth re-checking on what cadence.
What TAM Mapping Does Not Tell You
It sizes the opportunity; it does not validate it. A clean 8,000-account map for a product the market does not want is 8,000 accounts that will not reply, and the map will look correct throughout. It also says nothing about competitive density — a well-defined segment already served by three entrenched vendors is a harder market than a vague one that is empty.
Further Reading
Scoring prospects against your ICP
Finding decision-maker email addresses
The Bottom Line
TAM mapping is the step that converts a market opinion into an operational plan. Express the ICP as a filter, run it, read the count, tier the result, enrich only tier one, layer signal on top, and budget for decay. The count alone determines whether outbound is the right motion at all, which makes it the cheapest high-stakes number available to you.
The recurring mistake is treating the map as a one-time artefact. It is a query that should be re-runnable on demand, because the market moves, the data decays, and a map maintained by hand stops being maintained the first busy quarter. If you would rather have the mapping and the system built for you, book a meeting.
FAQ
How is TAM mapping different from market sizing?
Market sizing produces a number for a business case. TAM mapping produces a list of named, contactable accounts with the attributes needed to write to them. One supports a funding conversation, the other supports a campaign, and they are rarely the same number.
How many accounts should a TAM map contain?
For systematic outbound, roughly 1,000–20,000. Below 200 the economics favour account-based selling with human research per account. Above 20,000 the filter is usually too loose, and tightening it beats scaling into it.
How often should the map be rebuilt?
Refresh tier one quarterly and rebuild the whole map annually from the filter. At 25–30% annual decay, a two-year-old map has lost roughly half its contact accuracy, which shows up as a rising bounce rate long before anyone connects it to the map.
Do I need a paid data provider?
For a first count, no — most databases will run a filter and return a number on a trial. For an operational map with verified contacts you do, because the cost of sending to unverified addresses lands on your sender reputation, which is far more expensive than the data.
