TEMPLATE
AI quality evaluation plan
The plan that turns the requirement sheet into an evaluation. It names the method, the data, the people, the acceptance rule and the record for each characteristic.
| # | Section | What goes in it |
|---|---|---|
| 1 | Purpose and scope | Which system and which edition the evaluation is read against, and what the result will be used for. |
| 2 | Method per characteristic | The method, the tool, the dataset and the person for each characteristic in scope. |
| 3 | Data and provenance | The evaluation data, its source, its date and why it represents the production population. |
| 4 | Acceptance rule | The threshold and what happens when it is missed: remediate, restrict, or accept with a recorded reason. |
| 5 | Independence | Who runs the evaluation and who reviews it, and the separation between the two where the risk justifies it. |
| 6 | Record and retention | The result, the model version, the date and the retention period, linked to the traceability record. |
How to use it
Fill the sections in order. The scope decision comes before the requirement, because a characteristic that is out of scope needs a reason, not a blank. The evidence section comes before the owner, because a requirement whose evidence cannot be named is not yet a requirement. Keep the edition and the date on the cover.
The paid kit carries this outline as a workbook with the record's evidence and gaps beside each row, and with the owner, status and due fields ready to fill.
Buy the kit, USD 299 All templates