The ICP exercise most companies skip

Ask ten people at most companies to describe their ideal customer and you'll get ten different answers, all of them vague and most of them optimistic. "Mid-market companies that value quality." That's not an ideal customer profile. That's a horoscope. And a fuzzy ICP doesn't just cost you a slide — it quietly makes every downstream decision harder, from which leads to chase to what to build next.

The difference between who can buy and who should

Most ICPs describe who is capable of buying: the right size, the right industry, the right budget. That's necessary and nowhere near sufficient. The exercise companies skip is the harder one — identifying the accounts where you win fast, deliver real value, and keep the customer for years. Those are rarely the same as the ones who merely can afford you.

The data to answer this is already in your house. It's in your closed-won and closed-lost records, your churn list, and your happiest and angriest customers. The work is just looking honestly.

A practical way to run it

Pull your best customers — not biggest, best. The ones who renew, expand, refer, and don't drain your support team. Then find what they share that your mediocre customers don't. You're hunting for the non-obvious traits:

• What triggered them to look for a solution in the first place — the event, not the demographic.
• What they had already tried and abandoned before they found you.
• Who championed the purchase internally and what made that person successful afterward.
• What was true about their situation that made your product click immediately rather than slowly.

Your ICP isn't who you'd like to sell to. It's the pattern hiding in the customers who already proved you right.

The output should be sharp enough to disqualify. A good ICP tells your reps who to walk away from, not just who to pursue — and the walking away is where the value lives. Every hour spent on a plausible-but-wrong account is an hour stolen from a right one. If your profile never causes anyone to say no to a deal, it isn't doing its job. The test of a real ICP is whether a rep can look at an inbound lead and confidently decline it, not whether it makes a nice paragraph on your website.

Write it down, put it where reps see it, and revisit it every couple of quarters as you learn. Then watch what it fixes downstream: marketing targets better, sales cycles shorten because you're talking to people primed to buy, product hears a coherent signal, and customer success stops firefighting accounts that never should have closed. The ICP exercise feels like strategy theater until you do it properly. Done properly, it's the cheapest leverage you have.