ICP scoring: a fit score and a reason for every visitor

Tell WhoLayer who you sell to, and it scores everyone it identifies from 0 to 100, with a one-line reason like "VP Sales at a 210-person SaaS company; viewed pricing twice".

Set it up

1. Open Ideal customer in the dashboard.
2. Describe your best customers in one or two sentences, for example: "VP+ sales or marketing leaders at B2B SaaS companies with 50–1,000 employees in the US."
3. Save. A small AI model turns your sentence into scoring rules once. Check the rules it shows you in plain English, and edit any of them by hand.

How a score is worked out

Scoring uses those rules directly, with no AI per visitor, so it's instant and free. Each rule has a weight from 0 to 5:

• Job title keywords, seniority (for example VP and above), and department (people only; they don't apply to company-level matches).
• Industry and company size (employees).
• Country the visit came from.
• Buying signals: views of your key pages (like pricing), repeat visits, and how recently they came by.

The score is the weighted share of rules a visitor matches. When we don't know something, like a company's size, it counts as a partial match rather than a miss. Titles with words you mark as "never a fit" (like intern) are capped at 15. Open any person to see Why this score, rule by rule.

Visitors are scored when they're identified and again on every visit, so the score and reason are ready before an alert goes out. To score people identified before you set this up, or after you change the rules, use Score everyone now. Shoppers aren't scored.

Use it

• People and Companies: sort by ICP score and set a minimum score. A company's score is the best score among it and the people identified there.
• Destinations: set a Minimum ICP score on any Slack, Teams, webhook, HubSpot, Pipedrive, Attio, or Salesforce destination. People and companies without a score aren't sent there.
• Slack and Teams alerts show the score and reason, and so do the notes we add in HubSpot, Pipedrive, Attio, and Salesforce (no custom fields needed). Webhooks carry icp_score and icp_reason on visitor.identified and visitor.high_intent, and GET /v1/visitors returns them too. See the payload.

Teach it with feedback

On a person's page, mark them Good fit or Not ICP. Then use Re-tune with feedback on the Ideal customer page: we send your description, the current rules, and up to 50 of your most recent examples to the AI model once, and show you the new rules. Examples include job titles and company facts only, never names or contact details. You can re-tune 5 times a day, and rebuild from a new description 5 times a day.

Good to know

• It's on every plan and costs no credits.
• Scores never use personal traits like age or gender, and the model is told to ignore them if a description mentions them.
• A score helps you prioritize. It isn't a prediction that someone will buy.