You ran 12 expert interviews.How well did you actually run them?

How it worked

The pipeline is the boring part — that’s the point. Here’s how the recordings became a permanent site.

1. 1

   Recordings transcribed with speaker attribution and timestamps

   Every score in the final report cites a transcript timestamp, so the consultant can listen back to the exact moment that earned — or missed — the score.

2. 2

   Rubric operationalized into observable behaviors

   She brought us 8 rubric dimensions. We broke each one into 4–6 observable behaviors — concrete, listen-able cues a reviewer could mark consistently. The rubric became a checklist instead of a vibe.

3. 3

   Every interview scored independently against the rubric

   Each score paired with timestamped evidence — at least one supporting quote, plus a counterexample where one existed. No score floats unsupported.

4. 4

   Adversarial QA pass to eliminate score drift

   In a fresh session, a second reviewer challenged every score above 4 and below 3 — "show me the evidence this isn't 3." This is the most dangerous failure mode for self-evaluation: telling the client what they want to hear. We catch it.

5. 5

   Patterns synthesized across the quarter

   A heatmap across all 12 interviews. Recurring strengths called out with examples; recurring gaps called out with timestamped exemplars; coaching recommendations tied to listenable moments. The kind of synthesis you can't see on yourself.