Your ICP is a hypothesis, not a fact

Somewhere in this month's planning documents there is a sentence like this: "Our ICP is B2B SaaS and services companies, 20 to 100 employees, selling to a VP of Operations." It reads like a fact because it is formatted like one, sitting in a strategy deck next to actual facts. It is not a fact. It is a hypothesis, assembled from a handful of closed deals, some pattern-matching, and at least one segment somebody senior simply likes. The market has not voted on it yet, and when it votes, part of the document usually loses.

The document meets the dial

The vote happens fast once real calling starts. Within a few hundred dials per segment, patterns emerge that no planning session could have produced. One segment answers and converses but never advances, because the pain is real and the budget is not. Another segment was supposed to be the core and turns out to be unreachable at the title you chose, though the same companies engage one level down. A third segment nobody argued for at kickoff keeps producing the best conversations on the sheet. Every outbound program we have run has some version of this story, and the teams that win are not the ones whose original document was right. They are the ones who noticed fastest which parts were wrong.

Running the profile as a test

Treating the ICP as a hypothesis changes the mechanics of the first quarter.

• Write it in claims, not descriptions. "Operations leaders at 20 to 100 person logistics companies will take a meeting about X" is testable. A paragraph of adjectives is not.
• Give each segment a fair sample. A segment judged on forty dials was not judged, it was mugged by variance. Set a minimum attempt count per segment before any verdict, and hold to it even when early returns are ugly.
• Score segments on the same sheet. Connect rate, conversation quality, and meeting outcomes per segment, side by side, the same discipline as getting a baseline before setting goals. Anecdotes about segments are how the favorite survives the evidence.
• Separate reachability from fit. A segment that never picks up might be a list problem or a timing problem. A segment that converses and never advances is a fit problem. The fixes are different and conflating them kills good segments.

What to do with the loser

When a segment fails its test, the move is to park it in writing: what was tried, what happened, what would have to change to revisit it. Not deleted, parked. Markets shift and a losing segment sometimes becomes a winner two quarters later. The freed attempts flow to the segments that earned them, which is the same concentration logic as the fifty-account tier one: outbound capacity is finite, and every dial pointed at a mirage is a dial taken from a market that answers.

This is also an argument for humility in January specifically. The profile written at kickoff two weeks ago is at its maximum confidence and minimum evidence, and the gap closes only through volume. Our first weeks with any client are structured around exactly this audit: the segments get scored against live dials, and the ICP document gets its first honest revision inside the first month. If your profile has never been through that test, book a strategy call and we will tell you which of its claims we would test first.