Ideal Customer Profile Template

Many profiles come out of a workshop where the team describes the customers it likes. Start from records instead. Your CRM knows who signed: HubSpot, which makes CRM software, lists a deal's stage, amount, close date and whether it was won or lost among its default deal properties¹.

Who stayed is another matter. HubSpot can record when a deal's recurring revenue stops, and why, but only on Sales Hub Enterprise or Service Hub Enterprise, once someone sets up that tracking. Otherwise, renewals, upgrades and cancellations live in your billing system. So pull two exports, plus any call notes:

• Closed deals, won and lost: account ID, outcome, amount, close date, lead source and every company detail you hold
• Billing history for each account: start date, starting amount, current amount and status
• Discovery notes or call transcripts, where the sharpest disqualifiers often turn up

Join the files on the account ID. Count rows that fail to match in either direction instead of dropping them, since a high count usually means the IDs differ between systems.

Pick a horizon that includes at least one renewal. On annual contracts, 24 months covers the first renewal and the year after it. Your cohort is the won accounts whose billing started at least 24 months ago.

Leave younger accounts out of every rate, and say how many there are. They haven't had time to leave, so counting them as retained flatters the result. Measure from the billing start date, not the close date. An account that signs in March and goes live in July has had four months less to leave than its close date suggests.

For each segment, put five numbers in one row, with the count beside every rate:

• Win rate: deals won over deals closed
• Logo retention: the share of the cohort still paying at the horizon
• Net revenue retention: what the cohort pays now against what it paid at the start
• Value per account: what the cohort paid over the horizon, per account
• Value per opportunity: win rate multiplied by value per account

Point sales time at the last one, because reps work opportunities, not accounts. Win rate and value per account can disagree. Venture capitalist David Skok gives an example in his guide to SaaS metrics: bigger customers are harder to sell to, but usually place bigger orders, and churn less frequently².

A profile that only names who to pursue gives a rep nothing to point at when a bad deal turns up. Disqualifiers fix that, and each needs a number. Start with a payback threshold: what an account must return within the horizon to cover the cost of winning and onboarding it. A unit economics model works out acquisition cost and payback.

Look for disqualifiers inside your best-looking segment. Split it by what buyers said in discovery, and test each split against retention. A strong headline rate is often two groups averaged together.

Sort what you find into hard criteria, which mean decline unless a sales lead records an exception, and soft ones, which mean selling on different terms, such as self-serve. Also record what you tested and ruled out, such as industry, so nobody proposes it again. Make the hard questions required fields in your discovery call template, and count the exceptions each quarter against how those accounts performed.

Turn the attributes that separated stayers from leavers into a fit score, using only ones a rep can check before the sale. Record where each value comes from and how many accounts have it. A hand-typed industry field is weaker evidence than a location count you can check on a website. Give each attribute a small whole-number weight, totaling 100, so anyone can see why an account scored 65.

Then test the score on accounts it has never seen. The documentation for scikit-learn, a widely used machine learning library, calls testing a model on the same data it learned from a methodological mistake³. Split by start date, not at random, since the score will judge future accounts: fit the weights on older accounts and score the newer ones.

In both groups, compare retention in the top-scoring quarter with the bottom quarter. Expect the gap to narrow on the newer accounts, as it does in the template's made-up example (71.3 points to 54.5). Trust the smaller gap.

Personas come last, and only for your primary segment. One buyer rarely decides alone: Forrester's 2026 report on business buying found that the typical buying decision now includes 13 internal stakeholders and nine external influencers⁴. Write separate briefs for the weekly user, the person who signs, and anyone who can stall the deal without joining a call.

Base each brief on interviews. Nielsen Norman Group, a user experience research firm, warns that personas created with no new research⁵ can be "an echo chamber for the team's incorrect assumptions." It suggests starting with five to 30 interviews. Each brief should cover:

• What the person is measured on at work
• What they did before buying, and what made them act
• What they need in their first 30 days
• Their own words, quoted from a real call

Add one anti-persona: the buyer who is decisive, pleasant to sell to and gone at renewal. Naming the pattern helps reps spot it on the first call, not in next year's churn report.

Next, write a positioning statement for the segment the profile names, and map who matters in each deal with a stakeholder and champion map. The template on this page works through each step for a made-up company, and Edit with AI joins your own exports in River.