When your ICP is ‘everyone’ — a 3-step narrowing exercise

The single most common thing I hear on a first call with a founder who isn’t getting traction from outbound: “honestly, anyone with [problem] is a fit for us.” Usually said with pride, as if wide-applicability was the same thing as a strong market position.

It’s not. It’s the #1 reason outbound campaigns stall. An ICP that says “everyone” produces a list of 50,000 companies, a spray-and-pray sequence, a 0.4% reply rate, and a founder convinced that outbound doesn’t work.

The fix is almost never “narrow harder” in the abstract. It’s a concrete, 3-step exercise that takes about 90 minutes and produces a named 1,000–1,500 row target list with the specific reason each company is on it. Done it 40+ times. Works every time, provided the founder actually commits to the cuts.

Why “everyone” is the wrong ICP even when it’s true

The annoying part about “our product works for anyone” is that it’s often literally true. A good payroll tool works for any company with employees. A good observability tool works for any engineering team. A good email verifier works for anyone sending email.

But “works for anyone” and “worth cold-emailing anyone” are completely different questions. Cold outbound needs three things simultaneously:

• A recognizable trigger — a reason to email this company now instead of next quarter.
• A specific pain they’d nod at — so the first 30 words of the email prove relevance.
• A short proof path — so the meeting-to-pilot conversion isn’t a 6-month sell.

“Everyone with employees needs payroll” fails on all three at once. There’s no trigger specific to “having employees.” The pain is generic. The proof path depends entirely on what business they’re in. So the email ends up being “hi, do you have payroll needs?” — and deserves to be deleted.

Narrowing the ICP isn’t about excluding real customers. It’s about choosing the slice where the three things above are all true this week. The rest of the market is still a customer — just not for outbound.

Step 1 — List your last 20 real customers and find the pattern (30 minutes)

Every founder skips this step and jumps to “who should we sell to?” Don’t. Start with who you already sold to. Past behaviour is more informative than aspirational positioning.

How to pull the list

If you have 20+ paying customers: take the last 20 closed-won, in the order they closed. No cherry-picking.

If you have fewer than 20: take all paying customers, plus the last 10 hand-raised inbound leads that said “we’d be a fit.” This is worse data (inbound is pre-filtered by your marketing), but better than nothing.

If you have fewer than 5 paying customers: this exercise is probably premature. You don’t have enough signal to narrow an ICP. Get to 5 real customers first via founder sales, then come back.

What to record per row

In a spreadsheet, one row per customer:

• Company name, size (headcount), industry, country.
• How they found you (inbound / referral / outbound / event / etc.).
• The specific trigger or moment they decided to evaluate.
• The specific problem they were solving (in their words from the sales call notes, not your marketing copy).
• Who on their team made the call to buy.
• What broke first if you never sold to them.

The last one — “what breaks first without you” — is the highest-signal column. It’s the closest you’ll get to the real job your product does.

Look for three patterns, not one

Most founders look at this list and try to find the pattern. Wrong instinct. Look for three independent patterns:

1. Size/stage cluster. Are most customers 20–80 employees? 200–500? Enterprise? A customer base that’s evenly distributed across size bands is a red flag — it usually means your product is generic, not universal.
2. Problem cluster. When you read the “specific problem” column, how many distinct problems are actually there? If 14 of 20 customers have a variant of the same problem, that’s your ICP problem. If all 20 customers have 20 different problems, you have a positioning issue, not an ICP issue.
3. Trigger cluster. What made them shop? Hiring growth? New compliance requirement? Product launch? Failed internal project? Leadership change? One or two triggers usually account for the majority of buys — those are your cold-outbound triggers.

The intersection of (dominant size) × (dominant problem) × (dominant trigger) is your initial ICP. Not your final one, but the starting cut.

What if there’s no pattern

If you look at the 20 rows and there are genuinely no patterns, two possibilities:

1. You’re so early that your customer base is a statistical fluke. Keep selling, do this exercise again at 40 customers.
2. Your customers aren’t all benefiting equally. Look at usage or renewal data. The pattern is probably there, hidden inside a subset of customers who actually got value.

In my experience option 2 is more common than founders want to admit.

Step 2 — Cut until it hurts (30 minutes)

Now the hard part. Take the initial ICP from step 1 and deliberately narrow it further until you have fewer than 2,000 companies in your TAM.

This will feel wrong. Do it anyway.

The three narrowing axes

• Geography. If you sell to North America, cut to one country first (usually US). Then cut to one or two states or metros if the business justifies it. Timezone mismatches kill outbound meeting rates.
• Size. Take your dominant size cluster from step 1 and cut 30% off each end. If step 1 said 50–500 employees, start with 100–300. You can widen later.
• Vertical. Pick the top 1–2 industries from step 1 by customer count. Drop everything else. If “fintech” is 9 of 20 customers and “retail” is 3, do just fintech first. Retail is later.

Three cuts × the initial ICP usually takes you from 50,000 theoretical companies to 1,200–3,000 addressable ones. That’s a list you can actually work.

The “whole month” test

Here’s the pressure-check I use. Ask yourself: if I had to spend the entire next month selling only to this narrowed ICP, could I fill my calendar?

• If the answer is “easily” — the ICP isn’t narrow enough. A narrower ICP would convert at a higher rate with the same volume.
• If the answer is “probably” — the ICP is about right.
• If the answer is “no, I’d run out of prospects” — the ICP is too narrow for the volume you need right now, but probably correct for a 6-month founder-led sales motion. Keep it this tight and add the adjacent segment in month 2.

Most founders I work with answer “easily” on the first cut, which means they didn’t cut hard enough.

What to do with the segments you cut

Not throw them away. Park them. Create a spreadsheet tab called “Segments — later.” Every time you cut a segment, move it there with three fields: name of segment, reason it was cut, condition under which it would be re-added.

Example:
– Segment: “EU fintech, 300+ employees”
– Reason cut: Timezone friction + we don’t have a customer there yet
– Re-add condition: After we close our first US fintech customer at 300+ and have a case study

This matters because founders get nervous about cutting markets. Parking with a named re-add condition makes the cut reversible, which makes it easier to commit to in the moment.

Step 3 — Validate against three outbound triggers (30 minutes)

The narrowed ICP only works if you can run outbound into it. That means every row needs to have at least one of three things happening at any given time. If you can’t find triggers across the ICP, you narrowed on the wrong axes.

The three triggers I test against

• Hiring trigger. Are companies in this ICP consistently posting relevant roles? (See Apify + LinkedIn hiring signals.) If you search “VP Sales” + your narrowed industry + your narrowed geography + your narrowed size band and get fewer than 30 results in the last 30 days, hiring signals won’t carry the volume.
• Product/announcement trigger. Is there a regular drumbeat of public announcements — funding, product launches, partnerships, executive hires — in this ICP? Check the last 60 days of industry press. If it’s quiet, your signal library is thin.
• Peer-evidence trigger. Do you have 2+ named customers inside the narrowed ICP that you can legitimately reference in outbound (see Signal-based opener library)? If not, the peer-evidence pattern doesn’t work, which cuts one of your three strongest openers.

If at least two of the three are available, you have a workable outbound ICP. One is borderline — you can run outbound but you’ll need extra creative on the copy side. Zero means you have to widen the ICP or pick a different narrowing.

The validation worksheet

A simple four-row worksheet to run per candidate ICP:

If the “Usable now?” column has two or three yeses, ship the ICP. If it has one yes, revise either the ICP or the go-to-market plan. If zero yeses, stop — this isn’t an outbound ICP, it’s a founder-network or paid-ads ICP.

A worked example

Founder comes in with: “we sell observability software to any engineering team.”

Step 1 — last 20 customers:
– 13 of 20 are 80–300 headcount SaaS companies
– 11 of 20 bought because of a production incident in the 90 days before evaluation
– Most were brought in by the VP Engineering, closed with the CTO

Step 2 — cut:
– Geography: US only for v1
– Size: 100–250 headcount
– Vertical: B2B SaaS only (dropped consumer SaaS, dropped fintech-which-happens-to-be-SaaS)
– Result: ~1,100 companies

Step 3 — trigger validation:
– Hiring: “Senior Site Reliability Engineer” + narrowed ICP = 180 postings in last 30 days → ✅
– Announcement: Funding, product launches, outage postmortems published → ✅
– Peer evidence: 4 namable customers inside narrowed ICP → ✅

Outcome: the founder went from a TAM of “50,000 companies with engineering teams” to a working list of 1,100 rows with three live triggers. Reply rate on the narrowed campaign landed at 3.2%, vs. 0.6% on the original “everyone” version. Positive-reply-to-meeting conversion roughly doubled.

Same product, same company, same week. The only variable was the ICP cut.

Common failure modes in this exercise

“But we have real customers in the segment I’d cut”

Yes. You will. Cutting the segment doesn’t mean firing them — it means not hunting for more of them this quarter. Your existing customer will renew fine.

The mistake is letting the existence of one outlier customer prevent you from focusing on the cluster. Every campaign with a 15-segment “ICP” underperforms a campaign with a 2-segment ICP and 7 parked segments, roughly always.

“We can’t narrow — our investors want us to target the whole market”

A narrow ICP for outbound is not a narrow ICP for the company. Your website, your paid ads, your conference presence — those can stay wide. Outbound is a specific motion that requires specific cuts. Explain the distinction to the investor; most will understand it once they hear the framing.

“We tried narrowing and it didn’t convert better”

In roughly every case I’ve seen this, one of two things was actually true:

1. The narrowing was theoretical, not based on customer data. Going from “any SaaS” to “finance SaaS” doesn’t help unless finance SaaS is where customers actually come from.
2. The copy wasn’t updated to match the narrowing. If the email still says “we help companies like yours scale” after narrowing, the narrowing was invisible to the reader.

Narrowing without updating the copy to use the narrowing is wasted effort.

“The three triggers aren’t really there, but I want to pick this ICP anyway”

Then pick it. But run the motion as founder-led sales (warm intros, events, LinkedIn), not cold outbound. Different playbook. Cold outbound without triggers is a 0.3% reply rate, regardless of how good the positioning is.

What comes after the narrowing

Once the ICP is narrowed and validated:

• Build the row-level enrichment for the list (see Clay waterfall setup for the enrichment mechanics).
• Pick 2–3 opener patterns that match the available triggers (see Signal-based opener library).
• Set up the sending infrastructure so deliverability doesn’t tax the narrow list you just built (see Google Postmaster audit).
• Build the reply ops rhythm so the meetings the ICP produces actually convert (see Reply ops playbook).

Each of those playbooks assumes the ICP is tight. If yours still says “everyone,” do this 90-minute exercise first, then come back.

The uncomfortable truth

The narrower the ICP, the better outbound performs. The narrower the ICP, the scarier it feels to the founder. Those two statements are both true, and most founders choose the second one over the first. It costs them months.

If a 90-minute exercise and a temporary scope-down gets you from 0.6% to 3% reply rate, it’s the cheapest growth move on the table. Do it.