AI assisted call scorecards wired to coaching (L3)
Every recorded call gets scored against your methodology automatically, scores flow into the coaching system, and managers coach the two lowest scoring behaviours per rep. The model scores, the manager decides.
The steps
- 01
Write a scoreable rubric
Tool: Manual
Six behaviours maximum, each with an observable definition. Anything a human cannot reliably score from a transcript should not be in the rubric. Owner: enablement plus sales leadership. DoD: two managers scoring the same call land within one point.
- 02
Calibrate the model against humans
Tool: Claude
Score fifty calls both ways and compare. Publish the agreement rate. A rubric line below seventy percent agreement gets rewritten or dropped before rollout. Pitfall: shipping scores nobody trusts, which poisons the whole programme. DoD: published agreement rate per rubric line.
- 03
Route scores into the coaching flow
Tool: Salesforce
Scores land on the rep record and roll up to a manager dashboard with the two weakest behaviours highlighted. No new tool, no new meeting. Owner: RevOps. DoD: managers see current scores without asking anyone.
- 04
Coach behaviour, never the score
Tool: Gong
Coaching conversations reference clips, not numbers. Scores route attention. Clips change behaviour. Owner: frontline managers. DoD: every coaching note cites a timestamped clip.
Tools in this playbook
- Manual
- Claude
- Salesforce
- Gong
Next playbooks
Unfamiliar terms are defined in the AI and Revenue Dictionary. Related frameworks live in the framework library.
