A few years ago I sat in a pipeline review where the sales team had closed 14 deals and lost 23. The managing director asked the obvious question: what did the 14 have in common?
It took us two weeks to answer. It should have taken two hours.
That exercise was the first ideal customer profile the company ever built. Before it, the team had a product, a website and a target market the size of a phone book. After it, they had a shortlist.
The conclusion first: an ideal customer profile is a decision document. It tells sales and marketing which customers to chase, which to walk away from, and why. Most B2B tech companies do not have one, and it is the cheapest fix in go-to-market.
What an ideal customer profile actually is
A persona is a fiction with a stock photo. An ideal customer profile is a filter.
It answers five questions.
- What do they look like? Industry, size, ownership, geography.
- What do they already use? The CRM, the stack, the incumbent they are unhappy with.
- What is happening to them right now? The trigger. A new regulation, a funding round, a system reaching end of life.
- What do they gain, and what do they fear? The outcome they want and the risk they are trying to avoid.
- Who decides? The buying group, not the person who returns your call.
If you cannot answer all five, you do not have an ICP. You have a wish list with a logo on it.
If you sell two very different things, you may need two profiles. What does not work is one profile that tries to cover both, because it ends up describing nobody.
The three mistakes
1. You build it in a room with no customer data.
Most ICPs are written by the leadership team on a whiteboard, and they describe the customers the leadership wishes it had. The fix is boring: take your last 20 deals, ten won and ten lost, and look for the pattern. The profile is already in your CRM. Nobody has read it.
2. You make it a wish list.
Every customer sounds ideal when you are writing marketing copy. A real profile has an exclusion list. “We do not sell to companies under 50 employees.” “We do not sell to companies already locked into a five-year contract.” If you cannot say no to someone, the profile is not doing its job.
3. You write it and shelve it.
An ICP that does not change the next deal desk meeting is a PDF nobody opens. It has to live where decisions are made: the lead qualification form, the campaign brief, the quarterly planning sheet.
How long should it be?
One page. Two at most.
If your ideal customer profile is a 40-slide deck, it is a strategy document and it will be treated like one: read once, then ignored. The test of a good ICP is that a new salesperson can read it in three minutes and tell you which of their own leads to drop.
The document should also be easy to argue with. If nobody has ever said “that is not our customer” in a meeting, the profile is not being used.
How to build one in a week
You do not need a consultant or a survey. Five days and a spreadsheet will do.
Day 1. Pull the data. Your CRM, your last 20 deals, your last 10 lost deals, and the notes from the sales calls that did not convert.
Day 2. Look for the pattern. Three columns: what the wins had in common, what the losses had in common, and what neither had.
Day 3. Find the trigger. What was happening in the customer’s world in the six months before they bought? That is the single most useful line in the whole document.
Day 4. Build the shortlist. Pick the three segments where you win most often and could win again. Three is enough. Ten is a shopping list.
Day 5. Test it. Take it to two salespeople and one friendly customer. Ask one question: does this describe the deals we actually close? If the answer is no, the profile is wrong, not the team.
Where it feeds
For sales, the ICP decides which leads get a call and which get a polite no. It also decides how a rep spends Monday morning. Without it, every lead looks the same.
For marketing, it decides the message, the channel and the content. If you are ever stuck for something to write about, write about the problem your ideal customer profile has on the day before they buy. That is the content that gets read.
For pricing, a customer who buys purely on price was never your ICP. If your pricing conversations are always about discount, the profile is too wide, not the product too expensive.
If you want the wider process, we wrote a practical guide to building a full GTM strategy: GTM strategy for B2B tech companies
The buying side has changed
One more reason the profile matters more now than it did five years ago. Buyers do their own research long before they talk to you, and the most trusted sources are their own prior experience and peer recommendations. Your collateral comes lower down the list. The B2B Buying Disconnect research has tracked this for years, and the psychology behind it is laid out in Sales Psychology in Action. The summary is uncomfortable for vendors: good marketing loses to the customer’s memory of the last implementation.
A profile built from real customer data is the defence. You sell to the people whose prior experience matches what you actually deliver, and you stop pitching everyone else.
Start with the leak
You do not need a perfect ICP to begin. Look at the last ten deals you lost. If the same reason appears three times, that is your first insight. Fix that one thing and you have done more for revenue than any ten-page plan.
What is the one thing your best customers all have in common?