Rebuild a lead score that sales actually trusts

Rebuild a lead score that sales actually trusts
The playbook for fixing a model nobody acts on.

1 — Pull every closed-won from the last 4 quarters and reverse-engineer their pre-MQL behavior.
Why: your score should be built from what converted, not what a 2019 workshop guessed.

2 — Separate fit signals from intent signals into two scores, not one blended number.
Why: a perfect-fit account doing nothing is a different play than a poor-fit account hand-raising.

3 — Cap any single behavior's contribution.
Why: one webinar shouldn't outweigh a demo request, but uncapped point stacking lets it.

4 — Sit with three reps and have them score 10 leads by gut, then compare to the model.
Why: where rep instinct and the model disagree is exactly where your weights are wrong.

That's the stack for this week. Forward to a teammate.