# What should the recruitment data procurement checklist require before signature?


**Use non-negotiable acceptance gates before weighted supplier scoring.** The following is this desk’s recommended control framework, not a claim that every item is legally mandated in every jurisdiction.

| Gate | Minimum acceptance evidence | Suggested owner |
|---|---|---|
| Rule applicability | Counsel-reviewed map of entities, people, jurisdictions and intended decisions. | Legal and football operations |
| Permitted use | Clause-linked approval for exports, models, external AI, group sharing and retention. | Procurement and legal |
| Data provenance | Source, date, identifier and correction route for sampled player claims. | Recruitment analytics |
| Human control | Demonstrated review, override and recorded reasoning. | Head of scouting |
| Change management | Model and rules versions, release notices and regression tests. | Analytics and IT |
| Rights handling | Tested access, correction, deletion and legal-hold workflow. | Privacy owner |
| Coverage | Competition-season evidence, missing-data reasons and authorised samples. | Recruitment lead |
| Exit | Successful export and documented treatment of retained evidence. | Procurement and IT |

### What should happen over the next 30 days?

**Days 1–5: establish the decision.** Name the renewal or purchase, budget owner and deadline. Inventory actual workflows and prevent new uses from being added informally during a trial.

**Days 6–15: request evidence.** Send the Recruitment Data Licence Stress Test and AI Scouting Report Evidence Check. Ask football operations for current league instructions and counsel for jurisdiction-specific requirements.

**Days 16–25: run an authorised trial.** Test one identity correction, one adverse recommendation, one model update, one rights request and one exit export. For NWSL buyers, use women’s competition data throughout.

**Days 26–30: decide and document.** Record retain, add, replace, consolidate, defer or buy nothing. Unresolved permission, provenance or mandatory-control failures should remain visible rather than disappear inside an average score.

The final decision memo should answer: **what genuinely new evidence does this supplier provide, can we use it as intended, and can we explain and preserve the decisions it supports?**
