How to build an ideal customer profile that actually converts
A practical framework for turning closed-won data into a filterable ICP you can search against in minutes.
July 28, 2026 · 7 min read · Maya Chen
Most ideal customer profiles fail because they describe a feeling instead of a filter. If your ICP cannot be expressed as industry, size, geography, revenue band and technology signals, it cannot be searched — and if it cannot be searched, it cannot be repeated.
Start with your last 40 closed-won accounts. Tag each one with the attributes you can actually query: industry, headcount band, revenue band, country and state, and the tools they run. Patterns appear fast, and they are usually narrower than your marketing site suggests.
Next, subtract. Look at churned and lost accounts and find the attributes they share. Those become exclusion filters. A good ICP is defined as much by who you refuse as by who you chase.
Finally, translate the profile into a saved search. In Biztep you can store that combination once, run it weekly, and only review the new companies that entered the segment — which turns prospecting from a project into a habit.