ICP: how to define a B2B ideal customer profile with data

The ICP, or ideal customer profile, is the description of the company that buys from yours most easily, pays close to list price, stays longer and buys again. The definition that serves the operation comes from the CRM: you separate the customers who closed faster, with higher deal size, smaller discounts, lower churn and greater expansion, and look for what they share in industry, company size, region and source. The ICP the executive team writes during a planning afternoon tends to describe the customer it would like to have, or the biggest logo in the portfolio, and the sales team agrees with the slide and keeps prospecting whoever it already prospected. If you lead marketing, sales or post-sale, the profile only starts to count when your team decides with it day to day, in the list it buys, the account it prioritizes and the customer that gets dedicated attention.

In the diagnostics I run, I ask for the ICP and usually get one of two answers. The first is a sentence like “mid-sized and large companies, in services and manufacturing, in the South and Southeast”, which covers half of the Brazilian market. The second is a list of reference customers, chosen by contract size. When I cross won deals with cancellations, I often find a third profile, of companies that close in half the time and renew without negotiating, which nobody in the room had mentioned.

What is an ICP (ideal customer profile)?

The ICP describes a company with attributes you can verify before the first meeting: industry, size in revenue or headcount, region, business model, operational maturity and, in some sales, the technology the account already uses. It exists to decide where the company spends commercial effort, so it has to exclude accounts. One test I use is to hand the ICP text to two reps with the same list of one hundred companies and compare what each one marks. If their choices diverge, the criteria still depend on the reader’s interpretation.

The profile also needs an economic result behind it. The account that closes in three weeks and cancels in the sixth month consumes sales, implementation and support without returning margin. I define the ideal customer by the five results revenue leadership tracks: speed to close, deal size, discount granted, retention and expansion, which together show who pays back from the first meeting to renewal.

What is the difference between ICP and persona?

ICP and persona answer different questions, and confusing them hurts marketing and sales at the same time. The ICP says which company is worth investing in. The persona describes the people inside that company who take part in the purchase, with role, responsibility, pain and what each one needs to hear to support the decision. In a B2B sale with a buying committee, an account inside the ICP has three or four personas, and a well-handled persona in an account outside the profile produces a good meeting and a bad deal.

Aspect	ICP	Persona
Unit	Company, the account	Person, the contact
Question it answers	Which account is worth the effort?	Whom to talk to and what to say?
Attributes	Industry, size, region, business model, maturity	Role, responsibility, pain, decision criteria, typical objection
Main source	Won and lost deals, active and churned customers in the CRM	Recorded calls, discovery notes, customer interviews
Where it lives in the CRM	Company properties and segmentation rules	Contact properties and role in the committee
Who approves	Marketing, sales and CS leaders, with the executive team	Marketing, validated with sales
Day-to-day use	Prospecting list, media audience, portfolio priority	Message, content, discovery script

Why does an ICP written in an executive meeting become opinion?

The executive team writes the ICP from memory and ambition. Memory brings the cases that marked the company, usually the largest or most recent contracts, and ambition brings the market the company wants to win next year. Both work as a starting point for strategy and work poorly as qualification criteria, because the largest contract and the best customer tend to be different accounts. In a survey on key account programs, Gartner reported that customers elevated to key account status because of their large spend are 51% less likely to increase that spend over time (Gartner, 2021).

The effect shows up in every team. Marketing buys media for the profile on the slide, sales returns part of the leads because it recognizes another profile in its own history, and post-sale receives every kind of customer and treats them all the same way. Meanwhile the experienced rep follows instinct, and that rep’s portfolio becomes the company’s real ICP without anyone having written it down.

How do you define the ICP with CRM data?

The analysis uses won and lost deals from the last 18 to 24 months, active and churned customers and revenue per customer, which can come from finance if the CRM does not record renewals and expansion. If industry and size are blank for most of the base, the first step is to enrich those fields, and the gap already says something about how the CRM is used. This is the process I run with leadership, with the operations or CRM person pulling the data.

1. Define the results that characterize an ideal customer

Before slicing the base, the leaders agree on how to measure each result, with the time window adjusted to the sales and contract cycle.

Result	How to calculate	Where the data is
Speed to close	Days between deal creation and win	Deal dates in the CRM
Deal size	Value of the first contract	Value of the won deal
Discount	Difference between list price and closed price	Discount field or attached proposal
Retention	Renewed or cancelled within the contract window	Post-sale CRM or finance
Expansion	Additional revenue after the first purchase	Upsell and cross-sell deals, billing

2. Slice the base by account attributes

With the results defined, you compare each group of customers on the five numbers. The cuts that reveal the most are industry, size, region and lead source, followed by cycle, deal size and retention bands, which show which combinations concentrate the best customers. Source describes the channel and helps marketing decide where to invest, so it enters the analysis even though it is not part of the final ICP text.

Cut	Question for the base	Signal worth attention
Industry	Which sectors close faster and renew more?	Sector with many wins and many cancellations
Size	In which revenue or headcount band do deal size and retention hold?	Large accounts with high discount and low expansion
Region	Where is the cycle shorter and the cost to serve within margin?	Region with good deal size and expensive implementation
Source	Which channels bring customers who stay?	Channel with high volume and first-year churn
Cycle	Which profile closes below the median number of days?	Short cycle concentrated in small contracts
Deal size and retention	Who pays more and renews without renegotiating?	High deal size held up only by discount

3. Cross attributes in combinations

A single attribute misleads. Manufacturing can show up as the best industry on average, pulled up by mid-sized companies that came through referrals, while the large ones in the same sector churn. Cross two or three attributes at a time and avoid conclusions about groups of fewer than fifteen or twenty customers, because one large contract distorts the average.

4. Read losses and cancellations

Deals lost for lack of fit and customers who cancelled in the first year show whom the company serves poorly, and they tend to reveal exclusion criteria nobody had written. If the loss reason is filled in as “price” or “other”, read the notes and calls before concluding.

5. Validate with the people who serve the customer

The data shows the pattern, and the people who sell and implement explain the cause. Take the profile that emerged from the base to reps and account managers and ask why that group closes faster and stays longer. The answers bring attributes the CRM does not yet record, such as whether the account has a process owner, and those attributes become new fields.

6. Write the ICP on one page

The final document holds fit and exclusion criteria, priority tiers and the CRM field that supports each criterion, and it has to pass the test of two reps with the same list.

Which fit and exclusion criteria should you write?

Fit criteria split into mandatory and weighted. Mandatory criteria rule the account out if missing, such as minimum size or a region you serve. Weighted criteria add points and sort accounts into tiers, which I call A, B and C: tier A gets active prospecting and leadership attention, B goes into campaigns and lighter cadences, and C gets nurturing only until it changes tier.

Exclusion criteria describe the account the company decides not to pursue, even when everything else fits. Common examples are the company that only buys through public tenders when your operation does not handle that process, the account that requires an integration your team does not deliver and the customer whose pain lies outside what the product solves.

How does the ICP feed marketing, sales and CS?

Marketing: segmentation and media

For marketing, the ICP defines lists, media audiences and the profile criterion for an MQL. The media analyst builds the audience from tier A attributes, the content team writes for the pains of the personas in those accounts, and the budget gets evaluated by the share of in-profile leads that become opportunities. The profile criterion goes into the SLA between marketing, sales and post-sale, and the list of rejection reasons shows marketing when the ICP and the media audience have drifted apart.

Sales: qualification and portfolio priority

For sales, the ICP decides which account the SDR and the rep spend time on. In lead qualification, profile fit comes before the conversation about need and budget, and the rep records in the CRM which criterion was confirmed and which had no evidence. The manager assigns the portfolio by priority tier, reviews out-of-profile deals with discounts above policy in the pipeline and decides whether they continue, with that profile’s churn history in hand.

Post-sale and CS: onboarding and expansion

For post-sale, the ICP guides how service is designed. A tier A account gets dedicated onboarding and an expansion plan, and an account that came in outside the profile gets risk monitoring from kickoff, because history says it churns more. CS returns to leadership the signals that anticipate B2B churn and the attributes of accounts that expand, which go into the next profile review.

How often should you review the ICP?

I recommend a review every six months, with marketing, sales and CS leaders and the operations person presenting the same cuts from the original analysis over the most recent six months. The central question is whether tier A accounts keep converting, closing and renewing better than the rest, and whether any profile outside the ICP has started to perform.

The review indicators go into the RevOps metrics matrix: share of pipeline inside the ICP, conversion by tier, cycle and discount by tier, churn and expansion by tier. If tier A does not stand out from the others for two semesters in a row, the problem may lie in the criteria or in how the team uses them, and leadership needs to look at the CRM records before rewriting the profile.

Why does the ICP only change the operation when the team learns to use it?

The ICP is a decision rule, and each person in the operation makes dozens of these decisions a week. The SDR chooses whom to call, the rep decides whether to advance an out-of-profile deal, the manager assigns the portfolio, the analyst builds the campaign audience and CS chooses where to spend follow-up time. If the document sits in a shared folder, each of these people keeps deciding by their own history, and the CRM keeps showing the same customer distribution as before. In a survey published by Gartner, 94% of tech CMOs said they needed to improve adoption of segmentation, ICPs and buyer personas in the following twelve months (Gartner, 2024).

Adoption depends on training leadership and the team on the company’s own data. The manager learns to read the pipeline by tier and to lead the conversation with the rep who insists on an out-of-profile account. The SDR practices qualification with real accounts from the base, the marketing analyst builds the audience and compares it with wins, and CS learns to classify the new customer and record the signals that feed the review. In an in-company training, I work through these cases with the leaders of all three teams in the same room, because a criterion that one team applies alone turns back into opinion at the first results meeting.

ICP and L&D: what HR needs to assess

How to recognize the need

Reps complaining about lead quality, marketing that can only describe the ideal customer by naming a logo and CS serving very different customers with the same process point to the same gap. Before treating the case as an individual performance problem, ask the leaders whether there is a written ICP with verifiable criteria and who measures its use.

Who takes part

The core is the marketing, sales and CS leadership, with the direct managers of SDRs, sales and onboarding, who apply the criteria day to day. The operations or CRM person takes part to turn criteria into fields and reports, and the executive team takes part in the opening and in validating the profile.

How to measure

Satisfaction with the training says little about the ICP. Before the program, record the share of pipeline inside the profile and conversion, discount and first-year churn by tier. Compare after one or two quarters, knowing that the effect on retention shows up in a window equal to the contract cycle.

If your executive team has an ICP in the strategic plan and the CRM shows reps, campaigns and onboarding working different profiles, that is the work of the in-company Revenue Operations training. I run the program on the base and funnel of your operation, with marketing, sales and CS leaders building the profile together, and the group leaves with the SLA between marketing, sales and CS written from the ICP and with the RevOps metrics matrix that tracks each tier in the CRM you already use.

Frequently asked questions

What is an ICP?

ICP stands for ideal customer profile: the company that buys faster, pays more, asks for less discount, stays longer and expands, described by verifiable attributes such as industry, size and region.

What is the difference between ICP and persona?

The ICP describes the company worth investing commercial effort in. The persona describes the people inside that company who take part in the purchase, with role, pain and decision criteria. You define the ICP first and the personas within it.

How do you define the ICP for a B2B company?

Compare customers from the last 18 to 24 months by industry, size, region and source, measuring cycle, deal size, discount, retention and expansion. Validate the pattern with sales and CS and write fit and exclusion criteria on one page.

Can a company with few customers define its ICP with data?

Yes, with more caution. With a small base, reading losses, cancellations and conversations with the best customers weighs more than the cuts. Treat the first ICP as a hypothesis and review it the following semester.

How often should the ICP be reviewed?

Every six months, with marketing, sales and CS leaders looking at conversion, cycle, discount, churn and expansion by tier. A new offer, a new market or a price change calls for an off-cycle review.

How do you train the team to use the ICP?

Train by role with cases from your own base: the SDR qualifies real accounts, the manager reviews the pipeline by tier, marketing builds audiences and CS classifies new customers. The leaders of all three teams take part together, and measurement compares the share of pipeline inside the profile before and after the program.