ICP Segmentation Criteria: A Practical Framework for B2B Teams (2026)

Most “ideal customer profile” documents are a single paragraph: company size, industry, maybe a revenue range. That’s not segmentation, it’s a guess dressed up as strategy. Real ICP segmentation criteria are specific enough that two people using them independently would sort the same 100 accounts into the same groups.

This guide covers the four criteria that actually predict fit, how to combine them into segments instead of one broad ICP, and the mistakes that quietly make most ICP definitions useless the moment a rep tries to use them.

Why a Single ICP Usually Fails

A single ICP statement like “mid-market SaaS companies, 50 to 500 employees” describes thousands of companies with wildly different buying behavior. A 60-person bootstrapped company and a 450-person Series C company both fit that range, but they buy differently, on different timelines, with different people involved. Treating them as one segment means writing messaging too generic for either.

Segmentation fixes this by splitting your addressable market into groups you can actually write different plays for, not by finding one perfect definition.

The Four Criteria That Actually Predict Fit

1. Firmographic: What the Company Looks Like on Paper

Company size, industry, revenue band, geography, growth stage. This is the criteria everyone starts with, and the one that’s least predictive on its own. It filters out obviously wrong accounts but rarely tells you which remaining accounts will actually buy.

Use firmographic data to build the candidate list, not to make the final call.

2. Technographic: What They Already Run

The tools a company already has in its stack say more about fit than headcount does. A company running a modern CRM and a cold outreach tool is a very different prospect for a data enrichment product than one still working from spreadsheets, even at the identical employee count. Technographic signals also reveal gaps: a company using three point solutions that overlap with your product is a stronger signal than one using nothing at all. See what technographic data actually is and how it’s collected for how reliable each detection method is in practice.

3. Behavioral and Intent: What They’re Doing Right Now

Hiring for a role your product supports, a recent funding round, a leadership change, a public statement about a relevant initiative. These are timing signals, not fit signals. A perfect-fit account with zero active signal is a account to keep on a list; a good-fit account actively hiring five SDRs is an account to call today.

4. Authority: Who Inside the Company Can Actually Buy

The most commonly skipped criterion. An account can match every firmographic and technographic filter and still be a dead end if the person you’re reaching has no budget authority, no technical ownership, and no real influence on the decision. Segment by function and seniority band relative to the deal size, not by job title text: a “Head of Growth” at a 30-person company often holds more real authority than a “VP Marketing” at a 2,000-person one.

Building Segments, Not One ICP

Once you have data across all four criteria, group accounts into 2 to 4 segments based on how they cluster, not by manually assigning ranges. A practical way to do this:

1. Pull your best existing customers (highest retention, fastest close, lowest support burden) and look at where they actually cluster across firmographic and technographic criteria. This is your evidence, not a guess.
2. Name each cluster by what makes it distinct, not by size alone. “API-first technical buyers” and “team-led marketing buyers” are segments you can write different messaging for. “50 to 200 employees” is not.
3. Assign a different qualification bar per segment. A segment with strong intent signals available (job postings, funding data) can have a lower firmographic bar, because the timing signal compensates. A segment with weak signal availability needs a tighter firmographic filter to avoid wasting outreach.
4. Revisit quarterly, not annually. Technographic and behavioral signals shift faster than firmographic ones. A segmentation built on last year’s tech stack data is already stale.

From Segments to Named Accounts and Contacts

A segmentation framework tells you which accounts and which roles to target. It does not, by itself, give you a name, an email, or a phone number. That’s a separate, mechanical step: once a segment defines the firmographic and role criteria, a B2B enrichment tool converts that filter into an actual list of companies and the specific people who hold the authority criteria you defined.

If you already know which accounts and roles you’re after and need the contact details themselves, see how to find a company owner or decision-maker’s contact information for the search methods and verification steps. This guide covers the filter logic; that one covers turning a name into a verified email and mobile number.

Common Mistakes

• Segmenting only by company size. Size correlates with almost nothing about buying behavior on its own. Combine it with at least one other criterion before treating it as a real filter.
• Writing an ICP from aspiration, not evidence. “We want enterprise logos” is a goal, not a segment. Start from your actual best customers and adjust from there.
• No authority criterion. A segment with no defined decision-maker role is a segment reps will fill in incorrectly, one contact at a time.
• Too many segments to maintain. Beyond four, most teams stop actually using the distinctions and revert to treating every account the same.
• Never revisiting the definition. A segmentation frozen at launch stops matching reality within a couple of quarters as your product, market, and best customers change.

Conclusion

ICP segmentation criteria only work when they’re specific enough that two people would sort the same list the same way: firmographic to build the candidate pool, technographic and behavioral to rank it by fit and timing, and authority to make sure the account converts into a real conversation with someone who can say yes. Get the framework right first. Finding the actual contact once you know who you’re looking for is the easy part.

Start free, 100 emails or 3 phone numbers per month, no credit card required.