Scoring models

A scoring model is your definition of a good lead. It is a list of rules: company size, industry, location, pages visited, job title and more, each with a weight. Every new lead is checked against the rules and gets a score from 0 to 100. The score is the ICP score you see on every lead, and it is what sorts your best leads to the top.

1. Qualify → open a model (or + New model, or Create your first model).
2. Add or change rules: find a criterion, choose the values, set the weight from Penalize to Very important.
3. Test a website to see how any company would score, then Save and Close.

How a rule worksCopy Anchor Link

Pick a criterion (for example Industry), then decide what it does:

• Score it: leads that match gain points; the weight says how many. A negative weight (Penalize) takes points away.
• Only include: only leads that match are kept; everyone else is left out.
• Exclude: leads that match are left out before they come in (Exclude leads).

The Live impact panel on the right shows, for the last 30 days, how many leads were identified, how many passed your filters and how many were hot, so you see the effect before you save.

Warm and hotCopy Anchor Link

Scores fall into bands: below warm, warm and hot. Warm starts at 40 and hot at 70 unless you change it under Edit score bands (Reset to 40 / 70 brings the defaults back). Changing the bands only changes the labels; it never rescores leads. Your automations can act on the band, for example "post hot leads to Slack".

Draft, live, defaultCopy Anchor Link

• A draft is saved but scores nothing yet. Publish makes it live.
• The default model is the one whose score shows in the Leads table. Only a live model can be the default (⋯ → Make this the default).
• Archive stops a model scoring at once; you can Restore it. Duplicate creates a draft copy to experiment with.

New rules score new leads straight away. To apply them to the leads you already have, see Re-qualify existing leads.

Common questionsCopy Anchor Link

Do I need a scoring model to get leads? No. Visitors are identified and shown with or without one. A model adds the score that sorts them and that your automations can act on, and its Exclude rules keep out the leads you never want.

Does a low score delete the lead? No. A low score only means a low ICP score; the lead stays in your list and is still charged a credit. Leads are kept out only by an Exclude or Only include rule, or by the Lead quality bar on the Leads page (Remove or block leads), and those leads never use a credit.

Does qualifying cost credits? Rules cost nothing. AI prompts (questions the AI answers about each lead) use your plan's monthly AI qualification allowance, not your lead credits; a lead that already has every answer is rescored without any AI use.

What makes a lead hot? A score at or above your hot band (70 by default). Which rules get it there is up to you: a company in your target industry and size, on a pricing page, with a job title that buys from you, for example.

Can I have several models? Yes, as many as you like, each with its own rules: one per product line or market. The default model is the one the Leads table shows. Names must be unique on a website.

What is the ICP? Your ideal customer profile: who you sell to. In the dashboard you describe it with the rules of your scoring model; Ideal customer profile explains the short version your AI assistant reads.

How do I test a model? Click Test a website, type any company's domain and Score it to see how it would score. The calibration run scores 5 to 20 domains at once, good to check that your real customers come out hot.

Every control on this screenCopy Anchor Link

Every button, field and switch, with what it does.