An ideal customer profile (ICP) describes the companies that get the most value from what you sell and are worth the most to you: their industry and size, the event that starts their search, who decides, and what rules them out. Build it from won and lost deals, not opinions, then use it to choose accounts and ad targeting.
A common ICP is written in a meeting and fits almost everyone, such as mid-market companies in North America with 50 to 1,000 employees. That filter can’t pick accounts, set ad targeting or tell sales which lead to call first. The template below can.
What goes into an ICP template?
A usable ICP answers nine questions: two about the company, two about the moment it starts looking, and the rest about who decides, what they need to see, who isn’t a fit, how to spot a ready buyer and what the profile is built on.
| Field | The question it answers | Where the answer comes from |
|---|---|---|
| 1 Industry and segment | Where do our best customers operate? | Industry of won deals in the CRM, checked by hand |
| 2 Size and budget | How big are they, and how do they pay for this? | CRM and finance: headcount, sites or revenue, deal size |
| 3 Trigger | What event starts the search? | Customer interviews, notes from first sales calls |
| 4 Problem in their words | How do they describe the pain? | Call recordings, emails, interviews |
| 5 Buying committee | Who decides, who uses it, who can block it? | Contacts on won and lost deals |
| 6 Must-haves | What must a supplier prove to be chosen? | Win notes, reasons for losing, requests for proposal |
| 7 Disqualifiers | Who looks like a fit but isn’t, and why? | Lost deals, customers who left, deals with poor margin |
| 8 Signals | What shows a company is ready to buy now? | Job posts, permits, funding, events, website visits |
| 9 Evidence | What is this profile built on? | The number of won and lost deals and interviews, with dates |
Automation integrator selling to food and beverage plants
Built from 12 won and 5 lost deals · review in a year
- 1Industry and segment
US food and beverage plants: dairy, bakery, meat, ready meals. 50 to 500 people per plant.
→ Company list - 2Size and budget
Two or more production sites. A capital budget approved once a year.
→ Company list - 3Trigger
A new product line, a failed food safety audit, or a night shift they can’t staff.
→ Ads and pages - 4Problem in their words
We can’t staff the palletizing line. Changeovers take half a shift.
→ Ads and pages - 5Buying committee
The plant manager decides. A process engineer writes the spec. Food safety can block it. Finance signs.
→ Roles - 6Must-haves
Washdown-ready equipment, a fit with the lines they already run, service within a day’s drive.
→ Ads and pages - 7Disqualifiers
Single-site shops with no capital budget. Co-packers on short contracts. Buyers who want parts, not a project.
→ Exclusions - 8Signals
Job posts for automation technicians, an expansion permit, a visit to a packaging trade show.
→ Company list - 9Evidence
12 won and 5 lost deals from the last two years. Interviews with 5 customers and 2 lost buyers.
→ Backs up fields 1 to 8
Download the template as a CSV file, with blank fields and this example, for Google Sheets or Excel.
Two fields do most of the work. The trigger tells you when to show up, and the disqualifiers tell you where not to spend. Templates that stop at industry and size leave both out.
How do you build an ICP from won and lost deals?
Start from what already happened in your CRM, not from who you’d like to sell to. Lost deals matter as much as won ones: they show which traits only look like a fit.
Pull two years of closed deals
Export won and lost deals with industry, size, deal value, sales cycle, margin and how long customers stayed. If the list is short, use every deal and call the result a first draft.
Rank customers by value, not by count
Sort won customers by margin, retention and repeat orders. The top quarter is your starting profile. A segment with many small, slow deals can look big by count and still lose money.
Compare wins with losses
Check how often each trait appears in won deals and in lost ones. Keep the traits that are far more common among wins. A trait just as common among losses describes your market, not your fit, and the usual reasons you lose become disqualifiers. For example, if 9 of 12 won deals had two or more plants and 1 of 5 lost deals did, the number of sites belongs in the profile.
Talk to five customers and two lost buyers
Ask what started the search, who was involved, what nearly stopped the deal and what else they considered. Write the trigger and the problem in their words, not yours.
Test it on new deals for a quarter
Mark each new opportunity as a fit, a partial fit or not a fit. After a quarter, compare win rate, deal size and sales cycle across the groups. If fits don’t close better, rework the profile.
What is the difference between an ICP and a buyer persona?
An ICP describes a company. A buyer persona describes a person inside it. The ICP decides which accounts you go after. Personas decide what each person on the buying committee sees and hears.
| Compare | Ideal customer profile | Buyer persona |
|---|---|---|
| Describes | A company | A person on the buying committee |
| Built from | Won and lost deals, CRM data | Interviews, sales calls, support questions |
| Answers | Which accounts should we go after? | What does this person need to hear, and where? |
| Used for | Account lists, ad targeting, lead scoring | Messages, content, the sales conversation |
| In the example | Food and beverage plants with two or more sites | The process engineer who writes the spec and wants drawings, not slogans |
In B2B you need both, in that order. One ICP sits above several personas, one for each role that can say yes or no: here, the plant manager, the process engineer, the food safety lead and finance. Personas written first tend to describe the right people at companies that will never buy.
How do you turn an ICP into LinkedIn targeting?
On LinkedIn, an ICP maps almost field by field to targeting. Companies come from fields 1, 2 and 8, people from field 5 and exclusions from field 7.
- 1 · Profile
The ICP sheet
- Fields 1, 2, 8: who fits
- Field 5: who decides
- Field 7: who is out
- 2 · Company list
Named plants that fit
- Two or more sites, US
- Uploaded as a list of 300 to 300,000 companies
Left outSingle-site shops, co-packers
- 3 · Roles
The buying committee
- Operations · Manager to VP
- Engineering · Senior, Manager
- Finance · Director and up
Left outFunctions that never buy, entry level
- 4 · Ad set
List and roles together
- Company list + job function + seniority
- Audience expansion off
- Judged by target companies reached and sales conversations
Left outLook-alike reach outside the list
Build the company list
List the companies that fit by name and take out the disqualifiers before you upload. LinkedIn accepts lists of 300 to 300,000 companies. A list must match at least 300 members before an ad set can run, and LinkedIn recommends 1,000 companies or more (LinkedIn Help, checked 16 October 2026). If fewer than 300 fit, target by industry and company size, then check your named companies in the reports.
Add roles by job function and seniority
Turn the buying committee into job functions and seniority levels. LinkedIn maps each member’s job title to both with its own taxonomy (LinkedIn Help), so you reach people whose exact titles you wouldn’t think to list. In the example: Operations from Manager to VP, Engineering at Senior and Manager level, Finance from Director up. In LinkedIn’s list, Senior sits below Manager: it covers experienced specialists, not leaders (LinkedIn seniority codes).
Use job titles as a check, not the main filter. A plant manager at one company is a site lead at the next.
Turn disqualifiers into exclusions
Exclude the industries, company sizes and job functions you never sell to. LinkedIn drops anyone who matches an excluded attribute (LinkedIn Help). Audience expansion is on by default where available and shows ads to members similar to your audience (LinkedIn Help). For a named-company list, switch it off.
Expect a small audience
An ad set needs at least 300 members. LinkedIn suggests 50,000 or more in general, and 300,000 for Sponsored Content and Sponsored Messaging (LinkedIn Help). A tight ICP is often smaller, on purpose. Judge these campaigns by target companies reached and sales conversations started, not by cost per lead.
How does an ICP shape Google Ads?
Google search works the other way around. You can’t pick companies the way you do on LinkedIn. Microsoft Advertising comes closest: on search it can raise or lower bids by company, industry and job function from LinkedIn profiles, but it doesn’t limit who sees the ad (Microsoft Advertising Help). The search carries the intent, so the ICP decides which searches you pay for, which you block and what the ad says.
| ICP field | Where it goes in Google Ads | In the example |
|---|---|---|
| 3 Trigger and 4 Problem in their words | Keywords, ad text and the landing page | Searches like “palletizing robot integration” and “washdown conveyor system” |
| 6 Must-haves | Proof in the ad and on the page | Washdown-ready equipment that fits the lines you already run |
| 7 Disqualifiers | Negative keywords | jobs, salary, used, parts, manual, course |
| 2 Size and 5 Buying committee | One qualifying question on the form | How many production sites do you run? |
| 9 Evidence: current customers in your CRM | A Customer Match list to exclude current customers | Upload customer emails, then exclude the list |
Customer Match, Google’s way to use your own lists, works with people, not companies. It takes contact data such as emails, phone numbers and mailing addresses. Company names and domains aren’t among the identifiers it accepts (Google Ads Help). Targeting with those lists needs 90 days of account history and more than $50,000 in lifetime spend, but any account in good standing can use them to observe and exclude (Google Ads policy, both checked 16 October 2026).
So on Google, the ICP works through words: the searches you buy, the ones you block, and one question on the form that tells sales which leads fit.
What makes an ICP too vague to use?
Read each field and ask whether it would change a decision: which company to call, which ad set to run, which lead to drop. If it wouldn’t, rewrite it.
| Field | Too vague | Usable |
|---|---|---|
| Industry and segment | Manufacturing | Food and beverage plants with 50 to 500 people |
| Trigger | Wants to grow | A new line, a failed audit or a night shift they can’t staff |
| Buying committee | Decision-makers | Plant manager decides, process engineer writes the spec, finance signs |
| Disqualifiers | Left blank | Single-site shops with no capital budget |
| Evidence | Our team’s experience | 12 won and 5 lost deals, 7 interviews |
Questions about ideal customer profiles
What is an example of an ideal customer profile?
For an automation integrator: US food and beverage plants with two or more sites and a yearly capital budget, which start looking after a new line, a failed audit or a staffing gap. The plant manager decides with engineering, food safety and finance. Single-site shops are out.
Is an ICP the same as a target market?
No. The target market is every company that could buy from you. The ICP is the part of it that buys faster, pays more and stays longer. If it isn’t much smaller than the market, it can’t tell you where to spend.
How many ICPs should a B2B company have?
Usually one per offer with a different buyer. If two products are bought by different committees for different reasons, write two. If only company size differs, keep one profile with two size bands.
How often should you update your ICP?
Once a year, and sooner when win rates, deal sizes or your market change, or when you launch a new product. Rerun the comparison of won and lost deals on fresh data and date the evidence field.
Who should own the ICP, sales or marketing?
Both, with one person keeping it current. Sales knows why deals are won and lost, marketing turns the profile into lists and targeting, and the CRM owner keeps fields like industry and the reason a deal was lost clean.
Sources and review
- Template and example
- Ours. The example company is typical, not a client.
- Checked
- 16 October 2026. Platform rules change. The date shows when the sources were last read.
- What this guide does not claim
- That an ICP alone raises win rates or lowers lead costs.
- LinkedIn Help: Company list targeting
- LinkedIn Help: Target audience size best practices
- LinkedIn Help: Targeting options for LinkedIn Ads
- LinkedIn Help: AND-OR targeting and exclusions
- LinkedIn Help: Audience expansion
- Microsoft Learn: LinkedIn seniority codes
- Google Ads Help: Customer Match data formatting
- Google Ads policy: Customer Match
Sezemov, M. (2026). Ideal customer profile (ICP): a B2B template with examples. Wonderflow. wonderflow.agency/insights/ideal-customer-profile/
