How Do You Write an ICP Scoring Rubric in a Spreadsheet?

How do you turn your ICP into a score you can actually use?

Write the ICP as a short list of weighted criteria, give each criterion a clear point scale, and add them up in a spreadsheet so every account gets one number. A scoring rubric turns a vague description like "mid-market SaaS with a sales team" into a rule anyone on the team can apply the same way.

Most ideal customer profiles live in a slide deck. They describe the perfect buyer in a paragraph, everyone nods, and then reps go back to prospecting whoever looks interesting that day. The ICP never touches the account list, so it never changes behavior.

This tutorial is for a founder or a first GTM hire who has a CRM export and a spreadsheet, and wants a scoring rubric by the end of the afternoon. You do not need Clay, a data warehouse, or a scoring tool. Google Sheets or Excel is enough to start.

What is an ICP scoring rubric?

An ICP scoring rubric is a table that lists the traits of your best-fit accounts, the points each level of a trait earns, and how much each trait matters. Applied to an account list, it produces a fit score. The rubric is the written rule. The spreadsheet is just where you apply it at scale.

A rubric is different from a lead score. A lead score usually blends fit with behavior, such as email opens or website visits. A fit score only asks whether this account looks like the companies that buy, stay, and grow. I keep the two separate because behavior changes weekly, while fit should be stable.

It is also different from account tiering, though the two work together. Tiering groups accounts into A, B, and C buckets for effort allocation. The rubric is what produces the score that decides the tier. If you have already built a tier sheet, my guide on building an account tiering sheet from your CRM export pairs well with this one.

Which criteria belong in the rubric?

Pick four to six criteria you can actually find for most accounts and that clearly separate good customers from bad ones. Common choices are industry, company size, team structure, current tools, and a business model trait. If you cannot fill a criterion for most of the list, it does not belong in version one.

Start from your closed-won deals, not your opinions. Pull your last twenty or so customers and look at what they share. Then pull your churned customers and lost deals and look for what they share. The traits that appear in the first group and not the second are your criteria.

Write each criterion as something observable. "Has a dedicated revenue operations person" is observable. "Values data-driven decisions" is not. If two people would disagree about whether an account meets a criterion, rewrite it until they would not.

If your best customers split into two different shapes, do not force them into one rubric. Build one per segment. I covered that decision in how to pick an ICP when two segments both buy, and the short answer is to score them separately and compare.

How do you set up the spreadsheet?

Create two tabs. The first tab holds the rubric: one row per criterion, with columns for the criterion name, the weight, and the points for each level. The second tab holds your account list, with one column per criterion and a final column for the total score. Keep raw data and scoring separate.

On the rubric tab, give each criterion a weight that adds up to one hundred across all criteria. Industry might get thirty, company size twenty five, current tools twenty, team structure fifteen, and business model ten. The weights are your opinion about what matters, so write them down where everyone can see and challenge them.

For each criterion, define three levels: strong fit, partial fit, and no fit. Strong fit earns full points, partial earns half, no fit earns zero. Three levels is enough for version one. More levels create false precision and arguments about edge cases.

On the account tab, add one column per criterion where you record the level for each account, using a dropdown with the three values. Data validation in Google Sheets or Excel makes this easy and prevents typos that would break the formulas.

How do you write the scoring formula?

Convert each level to a multiplier, multiply by the criterion's weight, and sum across criteria. Strong fit becomes one, partial becomes half, no fit becomes zero. The total lands between zero and one hundred, which is easy for anyone to read and compare across accounts.

The simplest approach is one helper column per criterion. If the level column says strong, the helper shows the full weight. If it says partial, it shows half the weight. Otherwise it shows zero. A nested IF formula handles this, and a lookup against the rubric tab keeps the weights in one place so you can change them later without editing every formula.

Then add a total column that sums the helper columns. Once it works for one row, fill it down the whole list. Sort by the total, and you have a ranked account list built from a rule rather than from instinct.

If you prefer fewer columns, a single SUMPRODUCT across the level multipliers and the weights does the same job. I usually start with helper columns because they make it obvious why an account got its score, which matters when a rep asks.

How do you fill in the criteria for each account?

Use what your CRM already has, then fill gaps by hand for the top of the list. Industry and company size usually come from your CRM or an enrichment tool. Current tools and team structure often need a quick look at the company's website, job postings, or LinkedIn. Do the manual work only where it changes decisions.

Start by scoring every account with only the criteria you already have. That gives a rough ranking. Then do manual research on the top fifty or so, where a better score could change whether a rep spends time on the account. Leave the long tail with partial data until it matters.

If you later move this into a tool like Clay or HubSpot, the same rubric applies. The spreadsheet version is a prototype that proves the criteria work before you pay to automate them. Automating an untested rubric just produces wrong answers faster.

Mark unknown values clearly. I use a fourth option, unknown, that scores zero but shows up in a different color. That keeps an account from looking like a bad fit when it is really just under-researched.

How do you check whether the rubric works?

Score your existing customers and your lost deals with the rubric and compare. If customers score clearly higher than lost deals, the rubric is separating fit from non-fit. If the two groups score about the same, your criteria or weights are wrong, and you should revisit them before using the score on prospects.

This back test is the most important step in the tutorial, and the one most teams skip. It takes an hour and tells you whether your ICP is real or just a description you like. Look especially at the high-scoring lost deals. They often reveal a missing criterion, like a budget owner or a timing trigger.

Also look at low-scoring customers. If several good customers score poorly, the rubric may be too narrow. Either they are an emerging segment worth a second rubric, or a criterion is weighted too heavily.

Repeat the back test every quarter as you close new deals. Your closed-won notes are the best input for this, and I wrote about how to turn closed-won notes into an ICP update if you want a process for that review.

How should sales use the fit score?

Use it to decide effort, not to decide who is allowed to buy. High scores get personalized outbound and faster follow-up. Middle scores get lighter-touch sequences. Low scores get no outbound, but any inbound from them still gets a fair response. The score should guide where reps spend time, not replace their judgment.

Agree on thresholds with the sales team before you share the list. For example, seventy and above might be tier A, forty to sixty nine tier B, and below forty tier C. The exact cut points matter less than everyone using the same ones.

Put the score in the CRM as a property once it is stable, so reps see it on every account record. A score that lives only in a spreadsheet gets forgotten within a month.

What should you do next?

Export your account list, pull your last twenty customers and lost deals, and write four to six observable criteria with weights that sum to one hundred. Build the two-tab spreadsheet, score customers and lost deals first to back test the rubric, adjust, and only then score your prospect list and share thresholds with sales.

Keep version one simple. You will learn more from using a basic rubric for a month than from perfecting a complex one that nobody trusts.

If you want help turning your ICP into a scoring rubric, testing it against your deals, or moving it into HubSpot or Clay once it works, reach out. Let's chat about making your ICP something your team actually uses every day.