All notes
Builds 8 December 2025 1 min read

ICP scoring and TAM sizing solve different problems

Using your ICP score to size your TAM produces a number that's either too small or circular. They need separate methodologies.

ICP scoring tells you which accounts in your database are worth prioritising. TAM sizing tells you how many accounts like that exist in the world. These are different questions and they require different approaches.

The mistake: using the same firmographic criteria for both. You define your ICP as “SaaS companies, 50–500 employees, Series A–C, using Salesforce.” You run that filter on your CRM and get 800 accounts. You run it on a market database and get 12,000. You call that your TAM.

The problem: your ICP criteria are built from your existing customers, who are a biased sample. They’re the companies that already found you, evaluated you, and bought. The market contains thousands of companies with the same firmographic profile who have never heard of you, plus companies outside your current ICP who you could serve equally well if you’d ever pursued them.

A better approach for each:

ICP scoring should use behavioural and engagement data, not just firmographics. Firmographics tell you who could be a customer. Engagement tells you who is already in buying motion. The ICP score is a prioritisation tool for the pipeline you have, not a measure of the market you could address.

TAM sizing should start from the problem, not your existing customers. How many companies have the operational problem you solve? This number is often larger than your filtered CRM would suggest, because it includes verticals, geographies, and company profiles you’ve never actively pursued.

Keep these separate: a TAM number that gives you an honest picture of market opportunity, and an ICP score that helps you work through that market efficiently. One is strategy. The other is execution.

Use your ICP score to decide which 500 accounts to work this quarter. Use your TAM sizing to decide whether you’re in the right market at all.

Keep reading