Eighteen months after closing a celebrated Series A, a cybersecurity startup was staring at a net revenue retention of 74%.

The CEO thought it was a product problem. The VP of Sales thought it was a CS problem. The board thought both of them were wrong. There were calls. There were slides. There was a very expensive consultant.

But when the actual autopsy was done, the kind where you pull every account, overlay it against the real win/loss data, and talk honestly to the customers who left, the answer was hiding in plain sight.

They had 15 customers who were renewing, expanding, and referring. When you looked at them carefully, they all shared one specific situational trigger: every single one of them had been in the middle of a SOC 2 Type II compliance journey when they bought. The pressure of that deadline, the exposure it surfaced, the urgency it created, that was the match that lit the fire.

They had 60 other customers who looked identical on paper. Same company size. Same industry. Same buyer title. Same firmographics. But they hadn't been in that situational trigger. They'd bought because a rep was persuasive and the demo was good. And without the urgency, without the real pain, most of them never fully implemented, never got real value, and eventually churned.

The ICP document in Notion said: "Mid-market SaaS companies, 50–200 employees, US-based, Series A or beyond, Head of Engineering or CTO as primary buyer."

It said nothing about SOC 2. It said nothing about compliance inflection points. It said nothing about timing, triggers, or the specific moment in a company's journey where this product actually changed lives.

They had been selling to the wrong 60 companies for 18 months. Everyone had hit their targets. The pipeline looked great. The dashboard was green. And the business was quietly bleeding to death.

This is the ICP problem. Not a lack of a document. A document that described the companies they wanted to sell to, not the companies that actually needed them most.

The Ideal Customer Profile is the most discussed and least understood concept in go-to-market.

Every company has one. Almost nobody has it right.

Here's why: most ICPs are built backwards. They start from aspirations and work back to justification. The sales team wants to sell to enterprise. Marketing wants to target logos they can put on a slide deck. The CEO wants to raise the ACV. And so the ICP is designed to reflect those ambitions, rather than the evidence already sitting in the CRM.

The result is a document that describes a fictional version of your best customer, rather than the real one.

The real ICP is built forward from evidence. It starts with your actual best customers, not the largest, not the most recognisable, not the ones you're proudest of, but the ones who:

Then you work backwards. What do those companies have in common? Not just firmographically (industry, size, geography), but situationally.

This is the crucial distinction most companies miss. A robust ICP has at least three layers:

1. Firmographic fit: the visible attributes. Industry, company size, revenue band, geography, tech stack. These are necessary but not sufficient.

2. Situational fit: the invisible triggers. What is happening at that company, right now, that creates urgency and receptiveness? A compliance deadline. A failed implementation of a competitor. A new executive hire who wants to prove something. A regulatory change. A growth inflection that has broken the existing process.

3. Behavioural/psychological fit: how they buy and how they succeed. Do they have an internal champion with authority? Do they have the internal capability to implement? Do they move fast or slow? Are they data-driven or intuition-led?

An ICP that captures all three layers is a genuine targeting tool. An ICP that only captures the first layer is a demographic filter, useful, but catastrophically incomplete.

There is a related principle worth naming: "will buy" is not the same as "should buy." Any decent sales team can close accounts that don't belong in the pipeline. Persuasion, urgency creation, discounting, the tools of salescraft can get a deal over the line that should have been disqualified. But the cost of that close is paid in CS bandwidth, churn, poor references, and a diluted NRR that makes the next raise harder.

The job of the ICP is to define not just who will buy, but who should buy, whose life actually gets better because they bought.

Case Study: The Compliance Trigger

The cybersecurity startup above eventually ran a proper ICP rebuild. They brought together their top 5 CSMs, their 3 best AEs, and their Head of Marketing for two days. They pulled the entire customer database and sorted it by NRR, not by ACV. The bottom 40% by NRR, not by size, were examined in detail.

The pattern was unmistakable. The accounts without SOC 2 context had never operationalised the product. Not because of anything the CS team did or didn't do. But because without the compliance trigger, there was no internal urgency to do the hard work of embedding a new security workflow.

They rewrote the ICP. The firmographic layer stayed similar. But they added a mandatory situational layer: the prospect must be in an active SOC 2 journey or have recently completed one and be looking at what comes next. They added a qualification question to every discovery call. They built a sequence of event triggers in their CRM using job postings, LinkedIn signals, and Crunchbase funding rounds as proxies for compliance pressure.

Within two quarters, their qualified pipeline conversion rate went up 22%. Within four, their 12-month NRR crossed 110% for the first time.

Nothing changed about the product. Nothing changed about the sales process. They just stopped selling to people who weren't ready to succeed.

The ICP as a Shared Language

Here's the second failure mode: even when a company builds a decent ICP, it lives in a Google Doc that marketing wrote, that sales has never read, and that CS has translated into something entirely different.

I've walked into revenue rooms where the CMO's ICP, the CRO's ICP, and the VP of CS's ICP were three different documents describing three different types of companies. Marketing was targeting by industry and size. Sales was qualifying by deal size and decision-maker title. CS was segmenting by implementation complexity. All three were reasonable approaches. None of them were talking to each other.

The result? Marketing was sending MQLs that sales didn't want. Sales was closing deals that CS couldn't land. And CS was losing customers that had never been set up to succeed.

A shared ICP agreed, operationalised, and embedded across all three functions is one of the most valuable alignment tools a revenue organisation has. It is, effectively, the filter through which every GTM decision should pass.

People: Who Owns It?

The ICP should be co-owned by the CRO and CMO, with active input from the Head of CS. It should not be owned by marketing alone (they'll optimise it for targeting) or sales alone (they'll optimise it for closing) or CS alone (they'll optimise it for retention). Each function brings a different lens and the ICP is only complete when all three are in the room.

Assign a quarterly review owner, often RevOps, to stress-test the ICP against actual data. Markets change. Products change. Best-fit customers evolve.

Platform: Where It Lives

The ICP should not exist as a PDF or a slide. It should live in three operational places:

Tools like Apollo, Clay, and ZoomInfo can surface firmographic triggers at scale. 6sense and Bombora can layer in intent signals. But all of these are only useful if the ICP is specific enough to give them something to match against.

Process: The ICP Review Cycle