Lead scoring model design

When to use it

Sales complains about lead quality, marketing points at volume, and nobody trusts the current score — or there isn't one. Design the model before opening HubSpot's scoring tool.

The prompt

Design a lead scoring model for HubSpot. Context: we sell [WHAT YOU SELL] to [ICP]. Our best customers typically [SHARED TRAITS — industry, size, tech, behaviour before buying]. A lead is sales-ready when [YOUR DEFINITION]. Build: 1) A fit score (0–50) from firmographic properties we can actually populate, with point values. 2) An engagement score (0–50) from HubSpot-trackable behaviour, with point values and decay rules for stale activity. 3) The A1–C3 grid combining both, and which grid cells route to sales vs nurture. 4) Negative scoring — signals that should subtract points. 5) The 5 properties to clean or create first, since scoring fails on missing data. Challenge my assumptions where my ICP description and my sales-ready definition conflict.

How to use it

1. 1Fill in the placeholders honestly — the model is only as good as the ICP description.
2. 2Sanity-check the point values against 10 recent won deals and 10 recent lost ones.
3. 3Build the fit and engagement scores as two separate score properties in HubSpot.
4. 4Route only the top grid cells to sales for the first month, then widen.

Tips

• Decay rules matter more than point values — engagement without decay only ever grows.
• If a property is under 60% filled, don't score on it; fix the data or drop the signal.
• Revisit quarterly with fresh won/lost data.

Related resources