Data readiness — a workshop worksheet

Use this worksheet with the data owner, operational team and security representative before selecting a model.

Inventory the inputs

For each data source record owner, purpose, format, volume, refresh frequency and access route. Identify whether it is authoritative, copied or manually assembled. Note missing periods and known changes in collection practices.

Check identity and time

Verify that records can be linked to the correct customer, asset, document or event. Align timestamps and units. A model cannot reliably repair an evaluation built on mismatched identities or inconsistent measurements.

Assess permission and exposure

Have responsible teams confirm the proposed use, access boundaries and retention. Include derived datasets, logs and evaluation examples in that review. Use synthetic or appropriately prepared examples where live information is unnecessary.

Inspect quality with users

Sample ordinary and difficult records with the people who know the workflow. Identify whether missingness, labels or historical decisions reflect process problems. Document disputed labels rather than treating them as ground truth.

Produce a readiness decision

List blocking issues, remediation owners and a review date. A small, reliable dataset can support a bounded pilot; a large unreviewed collection does not establish readiness.

AIStrategy.gr working resource · 6 November 2026. Adapt this template to the actual workflow and responsible teams.