Predicted outcomes against actual outcomes, used to retune the scoring model.

What it holds

• Predicted score against achieved score for every decided bid.
• Bias by signal, showing which inputs over-predict.
• Model version history and the change made at each tune.

Sources. Outcome dashboard · Debrief records

• Bank freshness reportWhat is stale, who owns it and when it is due back.
• Cross-pursuit pattern libraryPatterns that repeat across bids, promoted into reusable content.