How to define an ideal customer profile you can build a list from

Define an ideal customer profile in two halves. The first is the firmographic half: industry, headcount, revenue band, location, tech stack, funding and open roles. Data tools sell every one of those facts, so have that half drafted and correct the draft. The second half is the one no database holds: two to five real customers and what they share, why they bought, the person whose work it changes, and who you will never approach. Write that half in sentences, turn the whole profile into the filters a search can run, and test it on conversations before you trust it.

Most of a profile can be looked up

Sort the usual profile fields by who can supply them and the list splits cleanly. Industry, headcount, revenue band, headquarters, founding date, tech stack, funding and news, open roles: each is a fact about a company that a data vendor sells, and some list tools will draft the whole set from your own website. None of it needs typing from memory. Let it be pre-filled and correct what is wrong, because confirming a draft is a much smaller job than facing an empty field.

Those fields are necessary, and a search cannot run without them. They describe the shape of a company, and a company of the right shape could want what you sell without being ready to buy it. Everything that decides whether it buys sits in the other half.

Start from two to five customers you already have

The input worth the most is a short list of your best current customers. Two to five is enough, and what they have in common becomes the filter set. It is the one input no vendor can fetch, because it is not a field on any record.

If you have no customers yet, say so in the profile. A profile is built from its example, and a made-up customer produces a made-up profile. One early profile said exactly that, and named the closest thing it had to a working pattern instead: the shape of the buyer's business.

Whichever you have, the property that separates your best customers from the rest is what goes in this part of the profile.

Write why they bought, as a mechanism

A list of traits tells a search what to match. The reasons tell you which matches are worth an email. Write them as mechanisms: what it is about a customer's business that makes your product pay for itself, or keeps it in place once it is there.

Here is how a profile for a product sold to B2B agencies, the kind that run lead generation and business development for their own clients, wrote its reasons:

• The best prospect builds the product into something it resells. An agency that bills clients for a retained business-development service cannot drop a tool inside that service without renegotiating its own client contracts.
• An agency compounds and a direct buyer is flat. One agency brings one client, then two or three more over time.
• Activation beats fit. A recent business-development hire, a new service line, a funding round or a public commitment to growth is a stronger signal than an industry match.
• Overlap in how they sell matters more than overlap in what they sell.

No search infers any of those lines. Each one also tells you what to look for on a company's site before it goes on the list. The first is the forever-customer test, and if you know the job buyers hire you for, it belongs here as well.

Name the person

The role you want to reach changes the whole people search. A company can have a head of partnerships and a head of procurement, and they are different searches with different results. Take the role from your reasons, never from a database.

The agency profile named the owner or the head of new business: someone whose own quarter depends on booking more first meetings. It also named who the person was not: a marketing manager or a procurement contact.

Say who it is not

Most profiles say only what to look for. Add the line that says never these.

A profile written for a headshot studio selling to startups listed its positive criteria as a growing team in the city centre, frequent hiring and founder-led branding. Its negative criteria were a mostly remote workforce and no public-facing brand. A mostly remote team is exactly the disqualifier that studio needs, and no search would infer it on its own.

Negative criteria come in two kinds. One is a standing rule, such as mostly remote. The other is a list of names: your existing customers, your competitors and every company you have already ruled out. A search will happily return all three, so name them in the profile and keep them on the do-not-contact list every new list is checked against.

The agency profile ruled out three kinds of company: agencies whose own product competed with the seller's, agencies with no named owner for new business, and any agency whose site showed no client work.

Turn it into filters a search can run

A list tool searches on structured filters, and it cannot search on a wish. Turning a dream customer into industry, headcount, geography, keywords and a signal is the job, and a list tool will not do it for you.

The firmographic half becomes the filters directly. The keywords and the signal come out of the reasons you wrote. The headshot studio's profile carried keywords such as founder photo, team portraits and hiring brand, and the agency profile's signals were its activation events. The rest of the profile (why they bought, the person, the exclusions) stays beside the list as what each row is read against before anything is sent.

Test it, and decide how you will read the test

A profile is a hypothesis until conversations bear it out. Record the segment of every conversation on the sheet you fill in after each call, so the profile can be checked against what people said.

Then be careful about what a small sample can tell you. A run of forty first calls, planned as about thirty with agencies and ten with in-house sales teams, meant to compare how often each group said yes to the one question that decided the call. The arithmetic showed it could not. With both rates near 40%, the 95% interval on the difference between a group of thirty and a group of ten is about plus or minus 35 points. Seven yeses out of ten in the smaller group, nearly double a passing rate in the larger one, would still read as noise.

So the plan read the split in one direction only: record it, keep the profile as written, and set down in advance how lopsided a result would have to be to overturn it. A tie would keep the profile too, because a tie argues against one particular cut and never for having no cut at all.

Five things

1. Let the firmographic half be drafted, and spend your time correcting it.
2. Start from two to five real customers and write what they share. If you have none, say so in the profile.
3. Write why they bought as mechanisms, and name the role from those reasons.
4. Add negative criteria: a standing rule, and a list of names.
5. Record the segment of every conversation, and decide before the results how you will read a small split.