ICP scoring: how to rank B2B accounts by fit, timing and reach

An ICP scoring model gives each company a number that reflects how promising it is right now, measured against your ideal customer profile. It is also called account scoring, because it works at the company level, which matches how B2B purchases are made: by a group, at a company, for a reason. Forrester’s State of Business Buying 2024 found an average of 13 people involved in a buying decision.

Scoring comes after qualification. Qualification is pass or fail: does the account meet the rules at all? Scoring ranks the accounts that passed, so a team with limited hours starts with the strongest.

• Key takeaway: score fit, timing and reach separately, then combine.
• Key takeaway: timing should decay. A signal at day 10 is worth more than the same signal at day 170.
• Key takeaway: keep the reasons next to the score, so a seller can see why an account ranks where it does.

ICP scoring rates companies; lead scoring rates people. Lead scoring usually adds points for engagement with marketing, such as form fills, email clicks and page visits, so it only works on people who already know you.

ICP scoring uses facts about the company and what is happening there, so it works before anyone has engaged. In B2B outbound, it comes first: it decides which companies to research and contact. Lead scoring becomes useful later, once people at those companies start to respond.

Fit-only scoring ranks companies that look like your customers, but it ranks them the same way every month. It cannot tell a ready account from one where nothing is happening.

Most scoring models are firmographic: industry, size and location add points. A perfect-fit company with nothing happening scores as high as one that just hired a new COO and announced an acquisition. Adding timing and reach fixes that. Timing says whether there is a reason to buy now, and reach says whether you can start a conversation with the people who decide.

Each component earns points up to its weight, and the three add up to a score out of 100.

Fit (up to 45 points)

A failed must rule or a matched exclusion disqualifies the account outright; it gets no score. Otherwise, verified must rules earn most of the fit points (55% of them in SalesOne’s default model), and preferred rules earn the rest in proportion to how many match. A must rule that could not be verified counts as half.

Timing (up to 40 points)

Timing starts from the strongest live signal. Its points depend on how strong that signal type is for you and where it sits in its window: full credit in the first half, less in the second half, and very little once it has expired. In SalesOne’s default model, one strong, fresh signal earns up to 30 of the 40 timing points, and each other strong signal type adds 4, up to the cap.

Reach (up to 15 points)

Reach counts the buying-committee roles found. Required roles weigh more than nice-to-have roles, and a role earns half credit when only a name is found without a confirmed current profile.

A scored account shows the points for each component and the evidence behind them. Here is a fictional mid-sized manufacturer scored against an ICP with three must rules, two preferred rules and four committee roles.

The same company a year earlier, with no live signal, would have scored 35 + 0 + 10 = 45: tier B, and below every account with a fresh, strong signal. Nothing about the company’s fit changed. The new COO and the hiring did, and the score moved with them. That is the point of scoring timing separately: it shows a seller when a familiar account becomes worth a call.

Set the threshold by capacity, not by a fixed number. A common approach is to work tier A first, then tier B, and to hold tier C until a new signal appears.

Because timing carries 40% of the score, an account with no live signal tops out at 60 and cannot reach tier A. That is intentional: fit says who to sell to, timing says when. If your team runs out of tier A accounts every week, widen the signal list or the windows before you lower the bar.

Start from your ICP rules, add signals and roles, pick weights, and then check the scores against real outcomes. Seven steps cover it.

1. Write your ICP as must, prefer and exclude rules.
2. Choose signal types and rate each one’s strength for your business: high, medium or low.
3. Set a window per signal type, and decide how credit falls off inside it.
4. Name the committee roles and how much each is needed.
5. Pick weights. Start with 45/40/15 and adjust only with evidence.
6. Show the reasons with the score: which rules passed, which signal, which roles.
7. After a quarter, compare scores with meetings held and deals won, and re-weight.

The common mistakes make scores either stale or unexplainable. Either way, sellers stop trusting them and go back to their own lists.

• Scoring on fit only, so timing never moves an account up and the same companies top the list every month.
• Counting signals without dates, which makes old news look fresh.
• Letting many weak signals add up to more than one strong one. Cap the bonus for extra signals.
• Hiding the reasons. Sellers ignore scores they cannot explain.
• Storing a score once instead of recalculating it as signals age.
• Never checking the score against outcomes.

SalesOne scores every qualified account on fit, timing and reach, weighted 45/40/15 by default, and places it in a tier. S1 deep research agents gather the evidence: the rules each account passed, the dated signal and its place in its window, and the committee roles found. The score is shown with those reasons, so a seller can check any number in a few seconds. Your ratings and reject reasons shape the next research run. S1 agents do the homework. Your team builds the relationship.