Most founders already have an ICP sentence. The problem isn't that they haven't written one. It's that they wrote it once, for a deck, and then used the same sentence to decide who to email, and those are two different jobs that pull in opposite directions.
The deck sentence is built to sound big
"We help revenue teams move faster" or "any company doing manual scheduling" reads fine on a slide, because a slide's job is to make an investor believe the ceiling is high. That's not dishonest. Breadth is the correct rhetorical move in a pitch, where the question being answered is "how large could this get," and a narrow, checkable sentence would undersell a real opportunity on purpose.
It just isn't the sentence you want open in front of you when you're deciding who to write to on a Tuesday morning, because it was never optimized to answer that question. It was optimized to answer a different one, for a different reader.
The outbound sentence is built to sound small
A targeting sentence that survives contact with a real inbox names a role, a size band, an industry, and often something that just happened: "engineering leads at Series A companies who shipped their first public API in the last two months." Read next to the deck line, that sounds almost embarrassingly narrow. It should. Narrow is the entire point, because the sentence's job here isn't to describe an opportunity, it's to describe a person you could write one specific, defensible email to this week.
A sentence that could describe five different job titles produces an email that has to speak vaguely enough to fit all five, which is exactly the email that gets skimmed and deleted. A sentence narrow enough to describe one produces an email specific enough to answer.
What happens when you paste the wrong one into outbound
Take the deck sentence and use it to decide who gets an email, and you'll write to people whose actual jobs, actual pressures, and actual reasons for caring are all slightly different from each other, using language general enough to not obviously misfire on any of them. It won't obviously misfire. It also won't land, because nothing in it could only have been written about the person reading it. That's the tell of outbound built from a fundraising sentence: technically accurate, addressed to no one in particular.
Write the second sentence on purpose, and expect to wince at how small it is
Take your deck sentence and ask a different question of it: which single person, this month, would I actually be glad to get on a call with? Write down the literal answer, not the category it belongs to. If the honest answer is "the head of ops at a twelve-person logistics startup that just raised a seed round," write that, titles and stage and all, and don't round it up to sound more like a market. It is doing a different job than the slide, and a good version of it will always look smaller than the number on your TAM slide, because it's describing thirty real people instead of a category that might contain thirty thousand.
Let the first replies edit the sentence, not just your instinct
The founder who wrote the sentence is guessing, however informed the guess is. The people who reply are not guessing; they're telling you, in their own words, whether the sentence described them accurately. After a small batch goes out, read the replies as edits to the sentence rather than as a scorecard on how the batch performed. A cluster of "I'm not the right person, try our ops lead" is the sentence telling you the title clause is wrong. A cluster of "how did you know we just did that" is the sentence telling you the trigger clause is exactly right and worth keeping.
Most founders treat the sentence as a decision made once, in private, before anything gets sent. It's closer to a draft that a handful of strangers get to mark up for you, for free, just by responding honestly. Rewriting it after fifty replies isn't admitting the first version was bad. It's the only way anyone ever gets a second version that's better.
Why founders resist writing the narrow one
Writing "thirty specific companies" down on paper feels like admitting the opportunity is smaller than the deck claims, and for a founder who just spent a month convincing investors otherwise, that feels like a step backward. It isn't. The deck sentence and the outbound sentence aren't in tension about the size of the opportunity; they're answering different questions for different readers on different days. An investor is being asked to believe in a category over five years. A prospect is being asked to reply to an email this week. Confusing which sentence answers which question is the actual mistake, not the size of either number.
There's a second, quieter reason founders avoid the narrow version: it puts a name on who might say no. A category is comfortable because rejection has nowhere to land. Thirty named companies means thirty specific silences if none of them reply, and that's a more personal outcome than a vague market simply not materializing. It's worth naming that discomfort honestly, because the alternative isn't avoiding rejection, it's avoiding finding out anything at all.
A short worked example
Say the deck sentence is "we help B2B startups manage customer onboarding." Pulled apart for outbound, a first pass might read: "founders at 5-to-20-person B2B startups who have hired a first customer success person in the last quarter." That's specific enough to write one real email to. Send to sixty companies matching it, and if half the replies say "we don't have a dedicated CS hire yet, our founder still does this," the sentence just told you the trigger clause is wrong and the segment is actually pre-hire, not post-hire. That's a better sentence than the one you started with, and it cost nothing but reading the replies honestly.
Keep both sentences, and know which one you're holding
You don't have to choose one sentence forever. Keep the deck version for the deck; it's doing real work there. Keep a separate, narrower, more specific version for outbound, and update it as replies come in. The mistake isn't having a broad sentence. It's not noticing you're holding it when you sit down to decide who gets an email today.
Turning that narrower sentence into an actual, named list is the part that used to take a spreadsheet and a weekend. Rocketship takes the plain-English version of it, shows you how many people it actually matches, and lets you read the next twenty-five by name before anything is added to your pipeline, so you can tell within minutes whether the sentence you just wrote describes thirty people or thirty thousand.