The Ideal Customer Profile, Written So You Can Actually Search For It

An ideal customer profile is only worth writing if it changes what you do on Monday. Most of them do not: they describe a company that "values innovation" and has "growth ambitions", and then the person building the list still has to decide, by feel, whether a particular twelve-person accountancy in Leeds counts.

A usable ICP is a filter. Every field in it has to be something you can check about a business before you contact it, from the outside, in under thirty seconds. That constraint is the whole method, and it rules out most of what people normally put in these documents.

ICP, persona and segment are three different things#

Worth separating, because the mix-up costs real time.

Term	What it describes	What it is for
Segment	A slice of the market — "dental clinics", "UK e-commerce"	Deciding which market to attack
ICP	The company you sell to	Deciding which businesses go on the list
Persona	The person inside it who signs	Deciding what the message says

Outbound needs all three, in that order, and the ICP is the one that decides who gets contacted at all. A persona document written before the ICP is a description of somebody imaginary.

The five fields, and why nothing else belongs#

1. Business type, stated narrowly enough to search#

Not "SMEs". Not "professional services". "Independent dental clinics", "single-site vets", "letting agents managing under 200 properties".

The test: could someone with no context find twenty examples using your words? "Dental clinics in Bristol" passes. "Ambitious healthcare businesses" does not.

Going narrow feels like leaving money on the table. It is the opposite: a narrow definition lets you write one first line that is true of every business on the list, which is what makes segment-level personalisation work at all.

2. Geography, at the granularity you can actually reference#

City, region or country — whichever level lets you say something specific. "UK" is fine for a SaaS with a remote audience. For anything sold to local businesses, city-level is the difference between a message that mentions a local landmark and one that could have been sent anywhere.

Geography also decides two practical things people forget: the language your emails and landing page need to be in, and which legal regime governs the sending. Cold B2B email is treated differently in Germany, in the US and in Canada, and the differences are large enough to change how you write.

3. A size signal that is visible from outside#

The single most common ICP failure is defining size by revenue or headcount, neither of which is visible for a private company on a Tuesday afternoon.

Use proxies you can actually see:

Visible proxy	Roughly indicates
Number of locations on the map listing	Multi-site operation, more process, slower decisions
Number of staff on the team page	Headcount, within about 30%
Number of reviews and their pace	Customer volume and how long they have been trading
Job adverts currently open	Growth and current budget
Whether they run ads on their own brand	Marketing maturity and existing spend
Website quality and CMS	Whether they have anyone doing this internally

Every one of those is checkable in seconds. That is what qualifies them for the profile.

4. A qualifying trigger#

The field that separates a list from a good list: what has to be true right now for this business to care?

Recently opened. Hiring for the role your product replaces. Running ads on a term they rank badly for. Just took over a second location. Has reviews mentioning the exact problem you solve. Has a booking form but no online payment.

Triggers decay, which is the point. A list built on a trigger has a shelf life measured in weeks, and it converts several times better than the same businesses contacted at random. If you cannot name a trigger, you do not have a timing argument, and your first line will have to work harder.

5. A disqualifier#

Write down what puts a business off the list. Franchises where marketing is decided centrally. Anyone under two years old if your pitch relies on existing volume. Businesses in a market you cannot service. Companies already using a competitor if you have no switching story.

Disqualifiers are worth more than they look. They are what stop a list of 400 from being a list of 260 real prospects and 140 wasted sends — and wasted sends cost you more than time, because bounces and complaints from badly matched contacts damage the sending domain itself.

Fill it in like this#

We sell to  [business type]
in          [geography]
that have   [visible size signal]
and         [trigger]
but not     [disqualifier]

The person who decides is  [role]
They care about            [outcome, in their words]
They currently do this by  [status quo you are replacing]

A completed example:

We sell to  independent physiotherapy clinics
in          Greater Manchester
that have   1–3 practitioners listed on the team page
and         take bookings through a form or phone only
but not     NHS-contracted practices or chains over 5 sites

The person who decides is  the owner-practitioner
They care about            not losing evening enquiries
They currently do this by  calling people back the next morning

Every line of that can be checked from a map listing and a website in under a minute — which means it can be checked at scale, and the list it produces is one where every business genuinely qualifies.

Building the profile when you already have customers#

Take your last twenty customers. Not the ones you wish you had — the ones who paid.

1. Sort by profit, not revenue. The customer who pays most is often the one who takes most support.
2. Take the top third and list what they share. Type, location, size proxy, what was happening when they bought.
3. Take the bottom third and list what they share. That list is your disqualifier field, and it is usually the more valuable of the two.
4. Ask three of the top third one question: what were they doing about this problem the week before they contacted you? The answer is your trigger, in their words, and it usually contradicts your marketing.

Twenty customers is enough. Patterns visible at twenty do not usually disappear at two hundred.

Building it when you have none#

You do not get to skip this step, and you do not get to guess your way through it either. You test it.

Write three candidate profiles — three different business types you have a plausible reason to serve. Build a list of thirty for each. Send the same offer to all three with only the first line changed. Two weeks later one of them has replies and two do not.

That costs ninety contacts and a fortnight, and it produces evidence rather than an opinion. It is also the reason to keep initial lists small: a first campaign is an experiment, and running it at 2,000 contacts means you have burned the market before learning anything. The mechanics of running it as a repeating cycle are in building a predictable pipeline for a small agency.

Turning the profile into a search#

This is the step that makes the document real. Each field maps to something you can actually query:

Field	Becomes
Business type + geography	The search itself — "physiotherapy clinics Manchester"
Size signal	A filter applied after collection, on review count, staff count, locations
Trigger	Either a separate search ("recently opened") or a check on each row
Disqualifier	A removal pass before anything is sent

Which source you search decides how much of that you can see. Map listings carry review counts, categories, opening dates and location counts, which covers most size and trigger signals for local businesses — the practical route from listing to contact detail is in finding business emails from Google Maps. Businesses with no premises, remote-first agencies and software companies come from web search instead. Creators and studios live on social platforms and rarely have a website at all — a different game entirely, covered in finding creators by niche.

Each of the industry pages on this site lists the searches that actually produce a clean list for that trade, along with which commercial accounts inside it buy repeatedly, which is usually where the profile ends up pointing after a couple of rounds.

Knowing when the profile is wrong#

The signals, in the order they usually appear:

• Replies say "we don't do that". The business type is too broad, or your disqualifier is missing.
• Replies say "wrong person". The ICP is right and the persona is wrong — you are contacting the practice manager when the owner decides, or the reverse.
• Interest, then silence at pricing. The size signal is off. You are talking to businesses too small to afford it.
• Good replies, slow deals. You have drifted up-market. Multi-site organisations reply politely and buy in six months.
• Nothing at all, from a verified list that delivered. The trigger is absent. Nobody is looking for this right now, and no wording fixes that.

Update the document when one of these shows up, and date the change. An ICP that has not moved in a year is either a very stable market or a document nobody is reading.

What to leave out#

Anything you cannot see. Budget, tooling, internal priorities, "values". Interesting, unverifiable, and therefore useless as a filter.

A revenue band you cannot check. Turnover figures from a data provider are stale, wrong for private companies, and expensive.

More than five fields. Every extra field shrinks the list without improving it. If you get to eight criteria and only forty businesses qualify nationally, you do not have an ICP, you have an account list — which is fine, but it wants a completely different approach.

Aspirational fits. The enterprise logo you would love does not belong in the profile until one of them has paid you.

FAQ#

What is an ideal customer profile?#

A description of the companies worth contacting, written so each criterion can be checked from outside before you make contact. It differs from a buyer persona, which describes the individual inside the company and shapes the message rather than the list.

How is an ICP different from a target market?#

A target market is a slice of the economy — "UK healthcare". An ICP narrows that to the businesses that specifically fit: "independent physiotherapy clinics in Greater Manchester with one to three practitioners". The first decides where to look; the second decides who goes on the list.

How many criteria should an ICP have?#

Five is a good ceiling: business type, geography, a visible size signal, a trigger and a disqualifier. More than that and the qualifying set gets too small to sustain outbound.

Can I build an ICP without any customers yet?#

Yes, by testing rather than guessing. Write three candidate profiles, build a list of thirty for each, send the same offer with only the opening line changed, and let two weeks of replies pick the winner.

How often should I revisit it?#

When the signals above show up, and otherwise once a quarter. Markets change slowly but your own capability changes quickly, and the profile usually needs updating because you can now serve someone you could not last year.