Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

B2B Marketing › B2B Demand

Building an Ideal Customer Profile From CRM Data

How to build an ideal customer profile from your own CRM: define a good customer, compare accounts that stayed with those that left, and write disqualifiers.
Tutorialintermediate14 min readUpdated

What a profile built from evidence changes

The ideal customer profile glossary entry defines a good customer in three parts: likely to buy, likely to succeed with what it buys, and likely to stay and grow. A profile written in a planning meeting usually captures only the first, because the accounts people remember are the ones that signed. Signing is the cheapest of the three to observe and the weakest guide to where the business makes money. An account that signs quickly and leaves at its first renewal costs its acquisition cost and returns little; an account that takes a long time to close and then spreads across several teams for years is the one the business should be looking for.

Building the ideal customer profile from the CRM changes three things. The definition of good becomes an outcome the business can measure, such as renewal, payback and expansion, rather than an opinion about which logos look impressive. The fields become traits that sales and marketing can check for any company before talking to it, so the profile can do work in targeting, scoring and qualification. And the disqualifiers become explicit: the traits that predict a poor outcome however keen the prospect seems. They are often the most useful lines in the whole profile, because they tell sellers what to walk away from.

Row labelProfile written from ambitionProfile built from CRM data
Starting pointTarget logos and the market the business would like to winAccounts that renewed, paid back and grew, compared with those that did not
Definition of goodLarge, well known, quick to signBought, succeeded and stayed, by measures agreed with finance and customer success
FieldsDescriptions such as "innovative mid-market firms"Traits checkable from outside: sector, size, systems in use, evidence of need, how buying is done
DisqualifiersRarely writtenThe traits that went with churn or a failure to pay back
UseA slide in the annual planFit points in scoring, routing rules and the target account list
ReviewWhen someone remembersA set date, once a new group of customers has reached renewal

The segmentation guide covers how a market is divided and which segments are chosen, and the STP framework is the sequence behind it. This tutorial is narrower: one business, its own customer records, and a profile written precisely enough that two people applying it to the same company would reach the same answer. The academy exercise on building a target account list for an industrial sensors maker practises reading a customer table that is handed to you; here the team builds the table first. The profile describes organisations, not the people inside them; the comparison of an ideal customer profile and a buyer persona sets out which decisions each document drives and how personas for the roles in matching accounts follow from it.

Before you start

  • Access to the CRM's account and opportunity records, including closed-lost opportunities and the dates each customer started, renewed, expanded or left. Someone who knows how the fields were filled in over the years is worth more than a tidy export.
  • Finance and customer success in the room for the first step. Finance holds revenue, margin and acquisition cost by customer, which payback needs; customer success knows which accounts are healthy, which are struggling and why.
  • A sales leader or senior seller who will sign the result off and can name the accounts they believe fit, as a check on the analysis rather than an input to it.
  • Win-loss notes and customer interview recordings, if they exist. The sales content and training guide explains how win-loss reviews are run; this tutorial uses what they produce.
  • A working file with no personal data. The analysis needs accounts and outcomes, not people. Leave contact names, email addresses and phone numbers out of the export; the guide to personal data in analytics explains the data minimisation principle behind that habit and how spreadsheets made for one meeting spread long after it.
  • Patience with small numbers. Many business-to-business firms have few enough customers that any comparison will be suggestive rather than conclusive. The steps are written with that in mind.

The steps

  1. Agree what a good customer means

    Before exporting anything, agree with finance and customer success how each of the glossary's three parts will be measured from the business's own records. Likely to buy is the easy part: the account became a customer, ideally without an unusually long cycle or heavy discounting. Likely to succeed usually means the product was adopted and the customer reached the outcome it bought for, which customer success can judge from usage and health records. Likely to stay and grow means renewed at least once, paid back its acquisition cost and, for the best accounts, expanded. Read payback on margin, as the retention and payback guide defines the payback period: the point at which the margin a customer has returned passes what it cost to win, not the point at which its revenue does. Write each definition down with the field it comes from, and agree what happens to customers too young to have reached a renewal: they stay out of the comparison and are checked against the finished profile later.

  2. Export accounts with their outcomes

    Export every account that reached an opportunity over a period long enough to include several renewal dates: won and lost, current customers and former ones. For each, carry the outcome fields agreed in the first step: won or lost, start date, renewed or not, churned and when, revenue and margin to date, acquisition cost where finance can attribute it, and expansion since the first contract. The churn and retention metrics guide sets out customer churn and revenue churn; use its definitions so that the outcome columns mean what the rest of the business thinks they mean. Then sort the accounts into groups: those that stayed and grew, those that stayed at their first size, those that churned or never paid back, and those lost before signing. Keep the file at account level, with contacts left out.

  3. Add the fields you can check for any company

    A profile is only useful if its fields can be checked for a company the business has never spoken to. Add those fields to every account in the export: sector, size by staff or revenue, region, ownership or structure where it matters, the systems in use that the product connects to or replaces, evidence of the need, and how purchases of this kind are made there, whether from a central budget, by a department head or through a procurement process. Fill them from the CRM where its data is reliable and from public sources where it is not, using the same rule for every account. Leave out fields that can only be known after a sales conversation, such as satisfaction with the current supplier; they belong in qualification, not in the profile. A field that is blank for most accounts in your own records is a warning that it cannot be used for targeting either.

  4. Compare the accounts that stayed and grew with the rest

    Lay the groups side by side, field by field. For each value of each field, count how many accounts in each group have it and what happened to them. Look for traits that are common among the accounts that stayed and grew and rare among those that churned or never paid back, and for traits that gather among the failures. Read the counts as the guide to reading small numbers advises: show the count behind every share, put a range around it rather than quoting a single figure, and claim no statistical significance from a few dozen accounts. A pattern that rests on a handful of accounts is a hypothesis to test with sellers and customer success, not a finding. Lifetime value (LTV) by group and the churn rate among accounts with each candidate trait are the two readings most worth putting in front of finance.

  5. Add the reasons from interviews and win-loss notes

    The numbers say which accounts did well; they rarely say why. The reasons sit in win-loss interviews, customer success notes and conversations with the customers who have stayed longest. Harvard Business Review's article on jobs to be done, listed in this page's sources, argues that customers hire a product to make progress in a particular circumstance, and that the circumstance explains a purchase better than the buyer's characteristics; the jobs-to-be-done framework sets out how to write that job down. Write the need in the profile the same way: the job the organisation hires the product to do and the situation that creates it, in the customers' own words. That is usually what separates two companies with identical firmographics, one of which stayed and one of which left. The win-loss section of the sales content guide covers running the reviews; here they are a source, not a project. Where the reasons are thin and new interviews are needed, the guide to researching business buyers covers reaching each role in the buying committee without letting account owners choose who is asked.

  6. Write the profile and its disqualifiers

    Write the profile as a short list of fields with the values that fit, each one checkable from outside: sector, size band, region, systems in use, evidence of the need, how buying is done, and the job in a sentence. Then write the disqualifiers, the traits that went with a poor outcome in the comparison: a size below which accounts never paid back, a structure in which nobody owns the problem, a commitment to a competing platform that the product cannot work alongside. Disqualifiers matter as much as the fit fields, because they give sellers permission to stop and give marketing a reason to exclude. Beside each line, note the evidence behind it and how confident the team is, so that the weaker lines are tested first at the review. Keep the whole profile short enough to be read aloud at the start of a sales meeting.

  7. Size the market the profile leaves

    Apply the profile to the market and count what is left. The market sizing guide defines the serviceable market as the total filtered by the segments, regions and sizes the business serves, and its bottom-up method counts the organisations that meet that definition from registers, directories or panel data. The count answers two questions. Is the profiled market large enough to carry the plan? And is it small enough for the sellers to work properly? A profile that leaves more accounts than the team could ever reach needs another field or a tighter band; one that leaves very few may have mistaken an accident of the sample for a rule, and its weakest line is the first to revisit. Write the count down with its method, because it is the baseline the review will use.

  8. Sign it off with sales

    Take the profile, the comparison and the count to the sales leader and the sellers who will use it. Their job at this step is to test it against the accounts they know: does it include the customers they would call ideal, and exclude the ones they wish they had never won? Where their judgement and the data disagree, look at the accounts in question together rather than settling it by seniority. Once agreed, the profile becomes one of the shared definitions in the sales and marketing alignment guide: segment, size, sector, problem and disqualifiers, with marketing targeting it, sales qualifying against it and both rejecting leads outside it. Record who signed it off and when.

  9. Put it to work in scoring, routing and the account list

    A profile that lives in a document changes nothing. Turn its fields into fit points in lead scoring, so that a lead from a matching company scores higher before it has done anything and a disqualifier removes points or keeps the lead from the sales stage; the lead scoring guide covers fit and intent signals, and the lead scoring model template has a fit table to fill in. Add the profile fields to the CRM as account fields that sellers and routing rules can read, so that matching accounts reach the right seller and qualification records the fit. Those rules only work if the values arrive in one form, and the tutorial on standardising CRM field values maps free-text sectors, countries and job titles to the profile's categories as records come in. Then apply the profile to named organisations to build the target account list, as the account-based marketing guide describes; the list, its tiers and the plays stay in that guide.

  10. Set the review date

    A profile is a hypothesis about which customers will do well, and every renewal tests it. Set a review date tied to the business's own renewal cycle, when a group of customers won under the new profile will have reached their first renewal and can be compared with those won before it. At the review, rerun the comparison, check the lines marked as weak, look at the accounts the profile excluded that were won anyway, and record what changed and why. New products, new markets and the last year's churn all move which accounts are ideal, which is why the glossary entry describes the profile as something revisited rather than set once.

Once the profile is agreed, the tutorial on piloting account-based marketing on a short list takes one segment of it into a first, deliberately small programme, and the demand section sets both beside the other ways of creating demand. Where the profile points to one industry rather than a short list of large firms, the comparison of vertical marketing and account-based marketing helps decide between building for the sector and working named accounts.

A facilities software firm's profile (illustrative)

Illustrative example — not a real result

Plinthwise Facilities Software compares renewals with churn

Plinthwise Facilities Software is an invented company selling maintenance-scheduling software to property businesses. Its old profile said "property companies in the region", which matched about 2,600 organisations, and its four sellers spent much of their time on small landlords who signed quickly and left at renewal.

Marketing, finance and customer success agree that a good customer has renewed at least once and paid back its acquisition cost on margin. Of 164 customers won over four years, 118 have reached a renewal date. The export groups them and adds the fields that can be checked from outside, and the comparison that matters is this one:

GroupAccountsSeveral sites, own maintenance teamSingle buildingOther
Renewed and expanded413335
Renewed at first size4726129
Churned or never paid back306195

The team reads it as a strong direction from small counts, not as proof. Interviews explain it: property managers running several sites with their own maintenance staff hire the product to keep planned maintenance on schedule across sites without chasing people by phone, while single-building landlords bought it for compliance reminders and left when they found a cheaper reminder tool.

The new profile names property managers with several sites and an in-house maintenance team, and adds two disqualifiers: a single building, and maintenance fully outsourced to a contractor that runs its own system. The "own maintenance team" line is marked as medium confidence, because it can only be checked from job adverts and company pages. Applied to the region, the profile leaves about 540 organisations, roughly 135 per seller, a list the sales leader agrees each seller can work. The profile becomes fit points in lead scoring, a required account field and the source of the target list, with a review set for 12 months later, when the first customers won under it reach renewal.

Verify it worked

  • Read your written definitions of bought, succeeded and stayed. Each should name the CRM or finance field it comes from.
  • Open the working file. It should hold accounts and outcomes only, with no contact names, email addresses or phone numbers.
  • Pick a company nobody has spoken to. Two people applying the profile from outside information alone should reach the same fit answer.
  • Look at the comparison table, as the illustrative example above does. Every share should sit beside its count, and each profile line should note its evidence and confidence.
  • Check the paper trail: the market count with its method, who signed off and when, the CRM fields routing can read, and the review date.

If it goes wrong

  • The profile leaves more accounts than sellers could reach. It is too broad. Add a field or tighten a size band.
  • The profile leaves very few accounts. It may treat an accident of the sample as a rule. Revisit the weakest line first.
  • Many fields are blank in the export. They cannot drive targeting. Fill them from public sources by one rule, or drop them.
  • The pattern rests on a handful of accounts. It is a hypothesis, not a finding. Show the counts and test the weakest lines with sellers and customer success.
  • Sellers reject the profile at sign-off. Their judgement and the data disagree. Review the disputed accounts together, not by seniority.

Next steps

Common mistakes

  • Starting from target logos. A profile built from the companies the business wants describes ambition and predicts little about who will stay.
  • Defining good as closed. Signing is one part of three; without renewal, payback and expansion in the definition, the profile favours accounts that buy quickly and leave.
  • Fields nobody can check. Traits that only appear in a sales conversation cannot drive targeting, scoring or routing.
  • No disqualifiers. Without them sellers have no permission to walk away, and marketing keeps buying leads from the accounts that churn.
  • Treating a handful of accounts as proof. Small counts give a direction; write the profile as a hypothesis and test its weakest lines first.
  • Contact details in the working file. The analysis is about accounts; personal data in a shared spreadsheet serves no purpose and travels a long way.
  • A profile marketing wrote alone. Unless sales sign it off as a shared definition, qualification and targeting keep using different ones.
  • No review date. Products, markets and churn change which accounts are ideal; a profile that is never rerun drifts away from the business.

Frequently asked questions

How many customers do we need before we can build an ideal customer profile from data?

There is no threshold that makes the analysis safe, and most business-to-business firms work with fewer accounts than any statistics textbook would like. What changes with the number is how the result is read. With few customers, the comparison between accounts that stayed and accounts that left gives a direction, written as a hypothesis with the count behind every share, and the interviews and sellers' judgement carry more of the weight. As more customers reach a renewal, each review tests the hypothesis again and the weaker lines are kept or dropped.

What should an ideal customer profile include?

Fields that can be checked for any company from the outside: sector, size, region, the systems it uses that the product connects to or replaces, evidence that it has the need, and how purchases of this kind are made there. Add the job the organisation hires the product to do, in the customers' own words, and a list of disqualifiers, the traits that predicted churn or a failure to pay back. Anything that can only be learned in a sales conversation belongs in qualification rather than in the profile.

Should the ideal customer profile include the companies we want to win but have not won yet?

Not as evidence. A profile built from target logos describes ambition, and it predicts little about which accounts will succeed and stay. Those companies can be tested against the profile like any other account: if they match the fields and none of the disqualifiers, they belong on the account list; if they do not, the profile is telling the team something worth hearing before sellers spend months on them.

Terms used on this page

Frameworks

Where this page is used

Also related

Sources

  1. Know Your Customers' Jobs to Be Done — Harvard Business Review

Related