Ideal customer profile template: Firmographics, fit signals and the closed-won analysis behind them

An ideal customer profile (ICP) template turns an account-level definition of your best-fit buyers into fields, a fit score per field, tiers and a target account list that sales and marketing share. It covers firmographics, business problem and impact potential, built from closed-won and closed-lost data. For example, 200 accounts with 0.25 clearing the fit threshold gives 50 fit accounts, and a 0.15 win rate gives 7.5 expected wins.

What is an ideal customer profile template

An ICP is an account-level definition of the companies most likely to buy from you, renew and expand. Altify defines it as "a working definition of the account most likely to buy a company’s product, renew it, and grow with it over time". It describes the company. A buyer persona describes a person inside it.

The template is what makes the definition usable. It turns the ICP into fields, a score for each field and a target account list that sales and marketing share. Without the list, the profile is a slide.

What the template holds

Altify says a complete ICP rests on three factors: firmographics, business problem, and impact potential. Firmographics are the easy part to fill. Crunchbase lists technographic data, funding or IPO status, company size, revenue, industry and location. The other two factors need judgment and interviews, and in our reading they are the fields that separate a fit list from a filtered database export.

ICP TEMPLATE (one row per account)
Account name:
Firmographics:   industry | employee size | revenue | location | funding stage | tech stack
Business problem: stated problem | evidence | source (interview, call, intent)
Impact potential: can the product move the metric it is bought for? yes / partial / no
Fit score:        firmographic points + problem points + impact points
Tier:             1 / 2 / 3
Last reviewed:    date

Where the fields come from

The inputs come from your own closed deals. Altify's first step is to pull a company’s closed-won and closed-lost history, and its last is to score new accounts against the pattern that separates the winners from the rest. Cognism suggests starting interviews with ten super users who share characteristics of your ideal customer profile.

One profile rarely covers every motion. Altify notes that most enterprise vendors run two or three variants tied to different products or motions, and that the ICP needs review immediately after a shift in win rate, deal size, or churn.

A worked example

These inputs are an example, not a benchmark. Say you score a list of 200 accounts and 0.25 of them clear the fit threshold: 200 accounts times 0.25 gives 50 fit accounts. In this example, at a 0.15 win rate, 50 fit accounts times 0.15 gives 7.5 expected wins.

That arithmetic is the reason the template exists. It ties the profile to a pipeline number you can check against your win rate benchmarks and feed into an account-based marketing program.

ICP vs buyer persona and why the template keeps them apart

The ICP and the buyer persona answer different questions, so the template gives each its own sheet. The ICP picks which accounts get pursued. The persona describes the people inside those accounts who buy, use, approve or block. As Total Product Marketing puts it, ICPs are focused on account fit and do not go into detail about the individuals in the buying group.

The comparison below follows the dimensions in NMS Consulting's ICP and persona guide.

Why the sheets stay separate

In our reading, the split protects the account score. An ICP score should rank a company the same way whether or not you have found its chief revenue officer yet. If persona fields sit inside the scoring model, accounts with thin contact data drop down the list for reasons that have nothing to do with fit.

The order of work matters too. NMS suggests teams build one ICP first, then create the minimum number of personas needed to explain materially different buying roles or decision patterns. HubSpot notes that most businesses maintain three to five core personas. We treat that as a ceiling to justify, not a target to fill.

Fields for the persona sheet

These fields come from the NMS persona template. None of them belong in the ICP score.

• Persona role: title family, responsibilities, reporting line, buying role, authority, budget influence and stakeholder relationships.
• Persona priorities: goals, key performance indicators (KPIs), current initiatives, urgent problems, personal exposure and internal commitments.
• Decision criteria: required capabilities, financial thresholds, risk conditions, implementation expectations, proof standards and procurement constraints.
• Buying role tag: whether the person is an economic buyer, champion, user, evaluator, approver, influencer or blocker.

A simple layout keeps the link between the two sheets explicit:

Sheet 1: ICP
  account_id | segment | fit_score | disqualifier_flag

Sheet 2: Personas
  persona_id | buying_role | priorities | decision_criteria | objections | proof_required

Sheet 3: Account contacts
  account_id | contact_id | persona_id

The account sheet decides who gets targeted. The contacts sheet maps known people to personas, so messaging can change by role while the account score stays fixed.

Which firmographic and fit fields belong in an ICP

The template sorts every candidate field into one of four groups. Context.dev names the four inputs of a useful ICP as firmographics, technographics, first-party behavior and situational triggers. HubSpot's worksheet describes the ICP as built on company level firmographic and technographic data points, such as employee size, technologies used and industry.

The test for any field is the same. Context.dev puts it plainly: a trait is useful only if it helps distinguish outcomes. A trait that nearly every company in your market shares is not a qualification rule. Run each candidate against your closed-won and closed-lost cohorts before it earns a row.

The four groups and how well a record holds them

The field names come from Context.dev's template. The reliability column is our reading, and it follows the decay rates that source describes: industry and business model may remain stable for years, while a website stack or leadership team can change in a quarter. Engagement signals can become irrelevant within days.

Why behavior and triggers sit apart from fit

Behavioral signals show interest, not necessarily fit. We keep them in the template because they feed the readiness score, not the fit score. The same applies to third-party intent data. Put disqualifiers in the firmographic and technographic groups as hard rules. Context.dev suggests explicit negative weights or hard exclusions for attributes that repeatedly produce failed implementations or churn.

What moves to the persona

Demographic and psychographic fields describe a person. Salesforce's customer profile guidance lists demographic information such as age, gender, location, and income level, as well as psychographic details like interests, values and lifestyle. None of these describes an account, so in our reading none of them can separate a winning account from a losing one. They go on the persona sheet.

One edge case: location. A buyer's home city is persona data. The account's billing region is firmographic geography, and it belongs in the ICP.

The ICP template to copy

The template below is our own layout. It takes the field groups covered above and turns them into four working pieces: an evidence table, a fit rubric, a disqualifier list and a tiered account view. Copy each into a spreadsheet tab or a custom object in your CRM.

1. Segment evidence table

One row per candidate segment. A segment earns a place in the rubric only when the closed-won and closed-lost columns disagree.

2. Fit scoring rubric

The fit score answers one question, as context.dev puts it: "How closely does the account match the durable characteristics of customers who succeed with your product?" Weights below are a starting example, not a benchmark. Set yours from the evidence table.

FIT SCORE (0 to 100), durable traits only
Criterion              Weight   Full match   Partial   No match
Industry or vertical     25        25           12         0
Company size band        20        20           10         0
Business model           20        20           10         0
Core infrastructure      20        20           10         0
Geography                15        15            7         0

Point bands
80 to 100  = Tier 1
60 to 79   = Tier 2
40 to 59   = Tier 3
below 40   = Not ICP
Any disqualifier = Not ICP, regardless of points

Readiness gets its own score, built from recent behavior and triggers. Keeping them apart stops what context.dev calls routing "a highly engaged but poor-fit lead as if it were an ideal account." Our lead scoring model template covers the readiness side, and the lead scoring glossary entry defines the terms.

3. Disqualifiers

These sit outside the points. context.dev's template lists example firmographic disqualifiers such as a services-only business or an unsupported region, and technical ones such as on-premises only or an unsupported framework.

4. Tiered account list layout

Account | Fit score | Fit tier | Disqualifier (Y/N) | Readiness score | Last trigger date | Owner | Next action

Sort by fit tier first, then readiness. A Tier 1 account with no readiness signal still gets outbound attention. For example, a list of 200 accounts with a 0.25 share in Tier 1 gives 50 accounts for named coverage.

Revisit weights on a schedule. HubSpot's guidance is to review the ICP every quarter using fresh CRM data, closed-won deal patterns, and updated customer feedback.

How to build an ICP from closed-won and closed-lost CRM data

The method runs in five steps. It uses data you already hold, then tests the weakest part of that data against buyers.

Step 1: Pull won and lost opportunities together

Export every opportunity that reached a decision. Clozd suggests starting with CRM data from the last 12 months. Keep closed-won and closed-lost in the same file. As Unkover puts it, "Studying only losses gives you a skewed picture."

Step 2: Cut by segment

Group the records by the firmographic fields from the template above: industry, employee band, region, tech stack. Keep each segment large enough to read, and merge thin ones rather than drawing conclusions from a handful of deals.

Step 3: Compute four metrics per segment

Unkover defines win rate as Wins / Total Decisions x 100. Use decided deals only, so open pipeline does not dilute the rate.

Segment win rate    = won / (won + lost) x 100
Segment cycle       = median(close date - open date), won deals only
Segment deal size   = median(first-year value), won deals only
Segment retention   = retained accounts / won accounts x 100

For example, and not as a benchmark: segment A has 200 decided deals at a 0.25 win rate, which gives 50 wins. In this example, segment B also has 200 decided deals but at a 0.15 win rate, which gives 30 wins. Before calling A the ICP, compare cycle length and retention. Compare cycles against our sales cycle length benchmarks and run both segments through the pipeline velocity calculator.

Step 4: Check the loss reasons with buyers

Do not take the CRM loss-reason field as fact. When Clozd compared CRM data from 1,000 closed-lost deals against first-hand buyer interviews, the information aligned only 15% of the time. If a segment looks weak because of "price," that label may be wrong.

Interview a sample from each segment. Unkover's guidance is to aim for at least 10 wins and 10 losses and to interview within 2-4 weeks of deal close. It also notes buyers respond at a 10-30% response rate, so build a pipeline of 50-60 deals to reach a full sample.

Step 5: Write the result into the template

Record the segment that leads on win rate, cycle and retention together as the core ICP. Record the segments that lose for fit reasons buyers confirmed as exclusions. Leave a loss reason off the template if only the CRM field supports it.

Worked example of segment scoring for a mid-market software seller

The scoring method below turns closed-won analysis into rubric weights. In this example, the inputs are round numbers, not benchmarks: 200 opportunities per segment and win rates of 0.25 and 0.15. We hold everything else equal, so the effect of fit on output stays visible.

Step 1: compute per-segment output

Pull each segment's opportunity count, win rate, median cycle days, average deal size and gross retention from the CRM. Then run two formulas per segment.

Pipeline velocity (per day) =
  (opportunities x win rate x average deal size) / median cycle days

Expected value per opportunity =
  win rate x average deal size x gross retention

Velocity tells you how fast a segment turns pipeline into revenue. Expected value per opportunity tells you what one more opportunity in that segment is worth once churn is counted. The pipeline velocity calculator runs the first formula if you want to check your spreadsheet.

Step 2: compare segments on the same base

In this example, at 200 opportunities and a 0.25 win rate, segment A yields 50 wins. In this example, at 200 opportunities and a 0.15 win rate, segment B yields 30 wins.

In this example, if cycle days, deal size and retention are equal across the two, that 20-win gap carries straight into velocity and expected value. In real data they rarely are equal. A segment with the lower win rate can still rank first on expected value if its deals are larger or it retains better, which is why the formula multiplies all three.

Step 3: set weights from the results

Rank segments by expected value per opportunity, then look at which firmographic fields separate the top segment from the rest. Those fields get the heaviest weight in the fit rubric. Fields that do not change between top and bottom segments get little or no weight.

Retention belongs in the weighting, not only win rate. HubSpot's guidance is to review your CRM for customers with the highest revenue, lowest churn, and strongest product adoption, and we read that as a case for scoring on value kept, not value signed.

Step 4: set tier cutoffs

Score every account in each segment with the new rubric. Place the Tier 1 cutoff at the score where the top segment's accounts cluster, and the Tier 2 cutoff where the next segment's accounts cluster. Accounts below both go to nurture or exclusion.

Then rerun the two formulas each quarter. If a Tier 2 segment's expected value per opportunity moves above Tier 1, swap the weights rather than defending the old rubric.

How narrow an ICP should be and how to turn it into a target account list

An ICP is narrower than the market you could serve. As context.dev puts it, your ICP identifies the subset most likely to become successful, profitable customers now. The working question is how narrow. We answer it from the capacity side: the profile should produce a list your reps can actually work, at the coverage your pipeline number needs.

Size the list against coverage and capacity

Start from the pipeline target, not from the database. Work back from bookings to the opportunities you need with the reverse funnel calculator, then check the ratio against plan with the pipeline coverage calculator.

Here is an example, not a benchmark. Suppose you take a draft list of 200 accounts. In this example, if 0.25 of them score into Tier 1, that is 50 Tier 1 accounts. In this example, if 0.15 of those 50 accounts close in the year, that is 7.5 wins.

In this example, if 7.5 wins does not cover the number, you have two levers: widen the profile or raise capacity on the accounts you have. In this example, if the 50 accounts exceed what your reps can plan for properly, tighten it. A broad list hides weak fit. As Altify notes, generic firmographic filters alone catch a broader, weaker list.

One profile does not have to cover every motion.

Test the draft on held-out accounts

Before you roll the profile out, hold back a slice of recent closed-won and closed-lost accounts that you did not use to build it. Score them blind. The test passes if Tier 1 holds a visibly higher share of wins than Tier 3. If the tiers look alike, the traits are describing your market, not your winners, and you go back to the cohort comparison.

Score it on the account record

A profile that sits in a document does not route anything. Altify notes that a profile scored against the Salesforce account record stays current, because it’s updated in the same place sellers already work. Write the score and tier to fields on the account object, then make assignment rules read those fields.

Tiers should decide coverage. Altify describes the ICP as the tool to decide which accounts get a dedicated rep and a full account plan. Rescore as deals close: the step most teams skip is scoring continuously as new deals close, not once a year.

ACCOUNT TIER RULES (fill in from your own scoring)
Field: icp_fit_score (0 to 100)
Field: icp_tier (1, 2, 3, Excluded)

Tier 1: score >= your cutoff
  Owner: named account rep
  Coverage: full account plan, multithreaded outreach
Tier 2: score between cutoff A and cutoff B
  Owner: pooled reps
  Coverage: sequenced outreach, plan on first opportunity
Tier 3: score < cutoff B
  Owner: marketing nurture
  Coverage: no rep time until fit or trigger changes
Excluded: any hard disqualifier = true
  Owner: none

Rescore trigger: every closed-won or closed-lost deal
Held-out check: Tier 1 win share vs Tier 3 win share

Agree the handoff for each tier in your marketing and sales SLA so a tier change moves the account the same day.

How often to update the ICP and the traps to avoid

An ICP drifts quietly. The product changes, a competitor reprices, a new segment starts closing, and the profile in the template stops describing who buys. The fix is a fixed cadence with two speeds.

The two-speed cadence

• Quarterly score check. Rescore open and recently closed accounts against the current ICP fields. Confirm that top-tier accounts still win and renew at a higher rate than the rest. This takes an afternoon and changes no definitions unless the gap closes.
• Full rerun of the segment analysis. Repeat the closed-won and closed-lost segmentation from earlier in this piece when win rate or retention in a tier moves outside your tolerance for two quarters running. Segment-level interview work needs a bigger sample: one win/loss guide puts it at 30-40 interviews per quarter when you cut by competitor, deal size or persona.

The same guide describes win/loss work as becoming cyclical once it's running, with each quarter building on the last. We think the ICP should sit on the same loop.

Traps that break the profile

1. Trusting CRM loss reasons. The dropdown records the rep's view. One analysis found that when buyers list several reasons, reps can accurately identify only one of those factors roughly 15% of the time. Rerun segments on outcomes and interview evidence, not on the loss-reason field.
2. Blending personas into the ICP. Job titles and pain points belong in the persona. Once they leak into the account profile, the score starts rewarding who you happened to reach rather than which company fits.
3. The slide-deck profile. If the ICP is not a set of fields on the account record, nobody scores against it, and drift goes unseen.

A dashboard check for drift

Add two tiles to your demand gen dashboard and review them in the quarterly check.

Tile 1: Off-profile share of new opportunities
  = opportunities from accounts below ICP threshold / all new opportunities
  Alert when share exceeds your ceiling

Tile 2: Tier win rate gap
  = win rate (ICP tier 1) - win rate (all other accounts)
  Alert when the gap narrows for two consecutive quarters

For example, and not as a benchmark: with 200 new opportunities in a quarter and an off-profile ceiling of 0.15, the alert fires above 30 off-profile opportunities. In this example, with the same 200 decided deals and a tier-one win rate of 0.25, you expect 50 tier-one wins. A shortfall that persists for two quarters is the trigger for a full rerun.

What this page does not cover

This page does not cover persona research methods in depth, the build of a readiness or lead scoring model, or the selection of data and enrichment tools. All numbers here are examples, not benchmarks, and nothing on this page is financial or legal advice.