# Research Brief

## Decision

- Decision owner:
- Decision to support:
- Action changed by the model:
- Decision date or frequency:

## Prediction target

- Target:
- Target definition:
- Prediction horizon:
- Information available at decision time:

## Success criteria

- Baseline:
- Primary decision metric:
- Required improvement:
- Operational constraint:
- Cost of a false positive:
- Cost of a false negative:

## Evidence boundary

- Data that will be used for training:
- Data that will be used for validation:
- Data reserved for final confirmation:
- Known limitations:

## Done when

- A non-technical stakeholder can identify the action and metric.
- The baseline and target cannot be confused.
- Every input can exist at the stated decision time.