Ideal Customer Profile Template: Build and Score Your ICP

Use this ideal customer profile template to define, score and validate target accounts, with steps, scoring weights and a tiered prioritization model.

The accounts that look best on paper are rarely the ones that buy first. That gap is structural rather than a failure of intuition. 72% of B2B purchases involve complex, multi-stakeholder buying groups, according to Demandbase's State of the B2B Buyer Report, which means a template built around one decision-maker describes a purchase that never happens.

A working template fixes the timing problem, not the description problem. Every row ties a durable company trait to a signal you can date, a source you can verify and a rule that tells you when to act. A company that scores poorly on fit today can become a priority next month because it raised a round or replaced its head of sales, and a template that can't express that shift is a filing cabinet, not a qualification system.

Table of Contents

The ICP Template Fields to Fill In First

A useful ICP template records fit, timing, evidence, and the rule for reviewing each field. Fill in the criteria before naming accounts, so the team can tell a durable match from a temporary buying opportunity.

Field

What to record

Example entry

Evidence source

Industry and sub-sector

Primary vertical and adjacent sectors

Logistics software, freight forwarding

CRM industry field, company registry

Employee count band

Headcount range your delivery model can serve

50 to 500 employees

Enrichment provider, company profile

Revenue or funding stage

Revenue band or latest funding round

Series A to B, USD 5M to 50M revenue

Funding databases, filings

Operating markets

Countries where you can sell and deliver

Singapore, Australia, UK

Company address, public filings

Technology stack

Tools your systems must connect with

HubSpot or Salesforce, no dedicated SDR team

Technographic enrichment

Buying trigger

Event that makes the account worth contacting now

New VP Sales hired in the last 90 days

Job posts, news, funding alerts

Disqualifiers

Conditions that rule an account out

Under 20 employees, no outbound budget

Win and loss notes

Deal economics

Expected contract value and sales cycle

USD 30,000 to 60,000 a year, 90 to 120 day cycle

Closed-won history

Buying roles

Economic buyer, champion and end user

CFO, Head of Growth, RevOps lead

Contact mapping in CRM

Evidence and freshness

Source, date, confidence, and review rule

Careers page, date checked, high confidence; recheck when hiring changes

Record in CRM with date stamp

Treat a missing source or review rule as an open question. Don't let an unverified field decide whether an account qualifies.

Three Layers That Must Stay Apart

A market segment groups companies with shared traits. An ICP defines the type of company worth pursuing within that segment. A target-account list names the companies that meet the criteria. The ICP framework for separating market segments, profiles, and account lists lays out these distinctions.

Keep those layers separate in your records. The ICP can specify that a logistics operator of a certain size typically involves a CFO, head of operations, and RevOps lead. The account record should identify the actual people in those roles.

That distinction also clarifies the difference between an ICP and a buyer persona. A persona describes a buyer; the ICP describes whether the buyer's company is a fit. A well-matched contact at a company that cannot deploy your offer is still a poor target.

The trigger and evidence fields add timing to the profile. Firmographic fit can stay stable while buying conditions change, so record when a signal was found and when to check it again. ICP insights from AI Optimization Services can serve as a related reading option while you decide which criteria belong in your template.

Missing firmographic or technographic values can be filled with enrichment from sources such as Getleads. Record the provider and check date alongside each value, then review time-sensitive signals before acting.

How to Build Your ICP From Closed-Won Data

A four-step infographic illustrating how to build an ideal customer profile using closed-won sales data.

Closed-won data shows which accounts bought. Closed-lost and churn data show where the apparent fit breaks down. Use all three to build an ICP that captures durable fit and the signals that tell you when an account is worth contacting.

The Seven Steps

  1. Export closed-won, closed-lost and churned accounts. Use a time window that captures your sales cycle, including deals that took longer to close. Pulling only large wins can make an outlier look like a repeatable pattern.

  2. Standardise the fields before comparing accounts. Define industry, employee count, revenue or funding stage, region, technology stack, deal size, cycle length, acquisition source and loss reason consistently. Reconcile differences such as currencies or competing headcount sources before analysis.

  3. Separate the accounts into winners, losses and churn. Comparing all three groups helps distinguish customers who bought and stayed from accounts that looked similar but did not close or did not retain. The ICP build guide from nagent also recommends including churned accounts.

  4. Compare traits across the groups and keep recurring patterns. A trait that appears across several winners and less often in churn may be useful. A technology stack found in only a couple of wins is weaker evidence; keep it as a hypothesis until you can validate it.

  5. Turn observations into criteria and record the evidence date. Make each criterion checkable, then mark whether it is supported by account records or remains an assumption. Add the source used to verify it, so an account's status can be revisited when its circumstances change.

  6. Set must-haves, exclusions and a timing trigger. Use the patterns in your data to select the conditions that make an account a fit and the signals that make it timely to contact. Funding, leadership changes, hiring activity and technology changes are possible signals, not automatic reasons to reach out.

  7. Keep the core compact. Start with a short set of firmographic and technographic criteria, explicit exclusions and a buying signal you can verify. A criterion that sales and data systems cannot apply consistently will not help prioritize accounts.

What a Finished Build Looks Like

The criteria should help explain why your best customers differ from churned accounts. If the groups look identical on paper, the profile describes a broad market rather than a useful fit.

Write the result in two forms: a plain-language summary for the team and structured filters or rules for your CRM and prospecting tools. Keep account fit separate from the roles involved in buying. A B2B buyer persona describes those people; the ICP defines the kinds of accounts to target.

For an ICP and list building engagement, The Social Search agrees the accounts, buyer roles and deal size worth targeting before building the list. This keeps the criteria tied to the buying moment, with fit, timing signals, verification sources and a review rule all visible in the finished template.

Scoring Weights That Make Your ICP Usable

"Mid-market" means something different to every rep on your team. "50 to 500 employees" means one thing, and a CRM can check it. Expressing criteria as testable thresholds instead of labels is the core recommendation in Graph8's ICP scoring guide, which also suggests weighting the criteria to a total of 100 points and giving the heaviest weights to the traits that separate your best accounts from the ones that churned.

The criteria fall into four families, firmographic, technographic, behavioural and operational, which is why a usable model carries more than company size and industry.

Criterion

Weight

Scoring rule

Example

Industry match

20

Full points for target sectors, half for adjacent

Logistics = 20, adjacent retail = 10

Employee count band

15

Full points inside the band, zero outside

200 employees inside 50 to 500 = 15

Revenue or funding stage

15

Full points inside the stage band

Series B inside band = 15

Operating markets

10

Full points for markets you can serve

Singapore = 10

Technology stack fit

15

Full points for compatible CRM and data tools

HubSpot = 15

Deal economics

10

Full points when value and cycle fit, partial near the edges

Contract near low end of target = 5

Buying trigger

15

Full points for a trigger within the last 90 days

New CFO hired 60 days ago = 15

Worked Example, Scored Line by Line

Take a retail distribution company with 200 employees, a recent Series B, operations in Singapore, HubSpot in place, a contract landing at USD 28,000 on a 110-day cycle and a new CFO hired 60 days ago.

Retail isn't the target sector, it's adjacent, so industry scores 10. Headcount of 200 sits inside the 50 to 500 band for 15, and Series B sits inside the funding band for another 15. Singapore is a market you can serve, so that's 10, and HubSpot is compatible for 15. The contract sits near the low end of target value, so deal economics earns partial credit of 5. The CFO hire is inside 90 days, so the trigger takes the full 15. Total: 10 + 15 + 15 + 10 + 15 + 5 + 15 = 85.

Run the arithmetic on a prospect from outside your target sectors and the same company with no trigger and the difference is obvious. That's the point. The score is only useful when it visibly moves.

Disqualifiers Override the Arithmetic

An account that scores 90 and trips a disqualifier is still excluded. Disqualifiers aren't negative weights that a strong industry match can offset, they're a hard floor, and the most common ones are unsupported technical requirements, an uneconomic service burden and a market you can't legally serve.

Store the weights and thresholds as CRM fields so any score can be audited back to its inputs. If you want the scoring to run without someone rebuilding a spreadsheet each month, automating lead scoring with CodeWords covers the mechanics. Pair that with segmentation rules your marketing team can name and the model stops living in one person's head.

Segmentation isn't decoration. A comparison of 18 million email recipients found campaigns aimed at defined segments produced a 14.3% higher open rate than broader targeting, as reported in OpenStax's chapter on B2B market segmentation. Weighted bands work the same way at the account level, provided you accept the limit of the finding: it came from one email comparison and doesn't prove segmentation alone causes the lift.

Before you score anything: write the weights down and date the document. A model tuned after a favourite prospect appears isn't a model, it's a justification.

How to Validate an ICP Before Scaling Outreach

Validation is two cheap checks and one small test, run in that order. Skipping to the test means you spend real sending volume learning that two people on your team read the same criteria differently.

Two analysts working together to evaluate prospect accounts and reconcile labels for customer profiling and outreach.

The Two Pre-Checks

Have two operators independently label the same 50 prospect accounts as fit or non-fit, then reconcile the disagreements and calculate how often they agreed. Artemis GTM's CRM data quality resource frames low agreement as a signal that the criteria are ambiguous or the data fields are missing, not that one operator is careless. If two experienced people disagree on a third of the list, a sequence will disagree on more.

The second check audits the records themselves. The same benchmark material puts typical CRM duplicate rates at roughly 10% to 30%, with best-in-class environments commonly cited below 5%, which makes deduplication and field validation a prerequisite rather than housekeeping. Data readiness is also a stated worry: in Validity's 2024 survey of CRM users, 67% said they were concerned about their data's readiness for artificial-intelligence and machine-learning applications (Validity). Automating a score on top of duplicates just makes the errors faster.

The Test Cohort

Pull 20 to 30 fresh accounts that satisfy the proposed profile and run them through a controlled sequence, separate from your existing customer-like cohort. Run the email side through Instantly so deliverability and sequencing are held constant, and the LinkedIn side through HeyReach so the social touches are logged the same way. Then compare reply, positive-response, meeting, opportunity and disqualification rates against the comparison cohort.

Break every result down by segment, trigger, market and data source instead of reading aggregate meetings. A high-fit account with no current trigger routinely underperforms a moderately fitting account in the middle of a change, and an aggregate number hides exactly that. The twenty questions worth answering before you commit to a target account are a useful check on which of those cuts is explaining the result and which one is noise.

The list itself needs to hold up too. We build and verify the list from trusted providers before a single sequence is switched on, because a clean segmentation model applied to scraped, unverified contacts measures the data vendor, not the ICP.

Set the Review Thresholds First

Decide in advance what would make you revise a criterion: frequent manual overrides, a material drop in positive-response rate for the test cohort, or an inability to verify the field consistently. Re-score the profile quarterly, and also after any change in pricing, product scope, market or win-loss mix. Keep version history so campaign performance can be tied to a specific template version rather than to "the ICP" in general.

Three Mistakes That Kill Validation

  • Measuring only meetings booked: a burst of low-quality meetings looks like success for a month and hides a broken criterion for a quarter.

  • Testing on accounts you've already warmed: inbound leads and conference contacts are pre-qualified by someone else, so they can't test your scoring.

  • Changing two criteria at once: you learn nothing about which one moved the result, and the next review starts from a mess.

How to Prioritize Accounts by Fit and Timing

A matched profile without a current buying moment stays low priority. An intent spike does not repair an account that fails the basic requirements. Flint's breakdown of buyer intent signal data treats those as separate dimensions: fit says the account could buy, and timing says it is likely to buy now.

A visual guide explaining how to prioritize business accounts using a fit and timing matrix strategy.

Signals Worth Recording

Funding, leadership change, hiring acceleration and technology replacement are the four triggers that most often justify contact this quarter, so record each one with its source, date, topic and confidence, and treat anything older than 90 days as stale.

Each row should hold durable fit next to that timing evidence. Source, date and confidence are the verification fields. The review rule is simple: a signal past 90 days drops the account out of the active queue until a newer one appears, even if the firmographic fit has not changed.

Third-party intent tools such as Whitewhale watch those four triggers between manual reviews, so the list can refresh without a full research cycle. Owned channels produce partial evidence too: social comments and direct messages tracked through Manychat, and WhatsApp replies routed with Respond IO. INFUSE's guide to intent activation notes that 61% of the buyer journey occurs outside observable channels, which is why confidence sits beside the signal instead of standing in for proof of readiness.

Tier

Score

Trigger status

Action

Tier 1

80 to 100

Active trigger in last 90 days

Personal outreach within 48 hours

Tier 2

60 to 79

No trigger, strong fit

Monthly nurture, monitor signals

Tier 3

40 to 59

Partial fit or stale trigger

Batch outreach only when a new signal appears

Tier 4

Below 40, or any disqualifier

Any

Exclude from outbound

Tier 1 gets a human-written first touch within two days, because the trigger ages faster than the fit. Tier 2 stays on monthly nurture while monitoring runs: strong fit with no change is a future customer, not a current one. Tier 3 gets batch outreach only after a new signal, and Tier 4 never enters a sequence.

A perfect-fit account with no trigger is a nurture record, not a priority.

The queue only works if the tiers are fixed before names are added. A target account list you can defend uses that same split of fit and timing. If you are buying contacts rather than building them, browse qualified leads only after those tier definitions exist. Contacts are scored and ranked on fit and intent separately, so a high-intent contact at a poor-fit account does not enter a sequence under the wrong label.

Confirm Your First Target Account List This Week

Seven days, two operators, one spreadsheet. Each day builds on the one before it, and nothing here needs a new tool.

  • Day 1: Fill the ten template fields for your best 20 customers and stamp every cell with its evidence date.

  • Day 2: Export closed-won, closed-lost and churned accounts, then confirm the criteria separate winners from churn. If they don't, the problem is the fields, not the list.

  • Day 3: Lock the 100-point weights in writing and score 50 candidate accounts against them.

  • Day 4: Have a second operator label the same 50 accounts and calculate agreement. Reconcile the gaps while they're fresh.

  • Day 5: Move the top 25 Tier 1 and Tier 2 accounts into a test cohort, recording the trigger and its source for each one.

  • Days 6 and 7: Compare the first results against your review thresholds and decide which criteria to keep, change or drop.

Store the template version and its date in the CRM, so the next quarterly re-score starts from a known baseline instead of someone's improved memory of the model.

Locked criteria also shape the inbound side. If buyers in your segments search before they reply, the same segment language should drive the pages they land on, which is why technical SEO work in SearchAtlas and content production through Outrank belong under the ICP rather than under keyword volume alone. Content assets and short-form video built with Narrareach carry the same segment message into channels where your trigger monitoring doesn't reach.

A usable ICP is one a team can verify, score and act on within a week.

The Social Search builds the profile and the list that follows from it, agreeing the accounts, buyer roles and deal size worth targeting before anything is scored. If you'd rather not start from a blank spreadsheet, we'll map your best-fit accounts, the buying signals worth tracking and a ranked build order into a free GTM Blueprint you keep either way. Visit The Social Search to request it, and a real person will reply within one business day.