The AI Call Scorecard Playbook
Turn Claude or ChatGPT into an honest coach that grades any call you make. Three copy-paste master prompts plus the 5-step setup roadmap, field-tested across seven weeks of real scored sales calls.
You can't improve what you don't measure — and for calls, measuring used to mean hiring a coach or guessing. This playbook gives you the complete DIY system: record the call, let AI build your custom scoring rubric, then grade every call against it with one non-negotiable rule — every score must cite a direct quote from the transcript.
It works for far more than sales. The same system scores internal meetings, vendor negotiations, client check-ins, even personal conversations. Anything you can transcribe, you can measure.
Everything in it comes straight from Ben's testing lab: seven weeks of scored sales calls, including one that went from a D to a B- in five days by fixing exactly what the previous scorecard flagged. Read the full experiment in the Brain Food article "I Built a System to Grade Calls With AI. My Average? Yikes."
What you'll learn
The three ingredients of call analysis: a transcript, a rubric, and an AI scorer with an evidence rule
The Rubric Builder prompt — AI interviews you and builds your custom scoring rubric in one conversation
The Call Scorer prompt — paste a transcript, get graded with your own words quoted back as evidence
The Trend Review prompt — spot the mistakes you repeat across calls and find the one fix with the highest payoff
The 5-step roadmap to go from zero to your first scored call this week
Starter rubric categories for five call types, including personal conversations