The short answer
To score leads with website intent, give more weight to depth and repetition than to any single visit. A visitor who returns three times, spends minutes on pricing, and brings colleagues from the same company scores high. A one-page bounce scores low. Build the score from the pattern, then route the high scorers to sales fast.
Weight the signals that predict buying
Start with the signals that actually predict buying: repeat visits, time on pricing and demo pages, several people from one company, and quick returns. Weight those. Then subtract for noise, like a single visit to one blog post.
Keep it simple. A score nobody understands gets ignored. A few weighted signals that map to real intent, refreshed often, beats a complex model that no one trusts.
The honest catch: a score is a priority cue, not a verdict. It tells you who to call first, not who will definitely buy. Treat a high score as a reason to reach out well, not a guarantee.
Frequently asked questions
What website behaviour should raise a lead score?
Repeat visits, time on pricing or demo pages, several people from one company, and fast return visits.
What should lower a score?
Shallow activity, like a single visit to one blog post with no return.
Is a lead score a prediction of a sale?
No. It is a priority cue for who to contact first, not a guarantee.
Turn website behaviour into a score that tells sales who to call first.