# Raw data vs recruitment models vs human scouting: when does each fit?


**Choose raw data when control and reproducibility are the priority; a modelled product when a useful analytical framework is available; and human services when the missing work requires assessment capacity or contextual investigation.** This is our decision framework, not a claim that one approach universally outperforms another.

TransferLab provides a concrete example of the modelling category: its documentation describes custom profiles and club-strength adjustments, while LCP explains its event-based modelling and API access. These descriptions establish functionality, not validated prediction quality for a particular MLS or NWSL recruitment pathway. ([analyticsfc.co.uk](https://analyticsfc.co.uk/transferlab/?utm_source=openai))

| Approach | Prefer it when… | Buyer retains responsibility for… | Ask for this evidence |
|---|---|---|---|
| Licensed feed plus internal models | The club has a defined question, engineering ownership and validation capacity | Model design, maintenance, interpretation and data governance | Authorised samples, schema, definitions, correction process and intended-use permissions |
| Recruitment analytics platform | Staff need repeatable profiles and comparisons without developing every component | Selecting appropriate reference groups and challenging recommendations | Model version, input lineage, sample size, missingness and relevant validation |
| Bespoke analytical advisory | A bounded decision needs specialist interpretation | Defining the question and acting on qualified conclusions | Reproducible calculations, assumptions, evidence trail and deliverable ownership |
| Regional or specialist scouting | The unresolved gap is observation, local knowledge or report production | Assignment quality, conflicts management and final football judgement | Named assessor, match references, dates, observation method and contrary evidence |

### How should league translation and role-fit claims be tested?

Ask the supplier to distinguish:

1. **Description:** what the player did in the observed context.
2. **Adjustment:** how the model changes that evidence for team or competition strength.
3. **Projection:** what the supplier expects in another context.
4. **Recommendation:** why that projection matters for this club’s role.

For a model that claims transferability, request a time-separated evaluation: predictions made using information available before the move, followed by outcomes measured afterwards. Ask how it handles players who receive few minutes, disappear from covered competitions or change roles.

Use destination-relevant samples. Require a clear statement when evidence is insufficient for women’s competitions, a particular league pathway or a small positional group. Do not accept a polished percentile chart as validation by itself.

### What should a regional scouting commission specify?

360 Scouting’s public services distinguish player evaluations, league scouting and scout placement. Do not infer live stadium attendance from those category names. ([360scouting.com](https://360scouting.com/for-clubs/))

Our recommended commission should state:

- The role and unresolved question.
- Required matches and whether observation is live or video-based.
- What the incumbent evidence already says.
- Report format, clip references and delivery deadline.
- Assessor identity and relevant experience.
- Agent, club and supplier relationships to disclose.
- Whether the work is exclusive, reusable or subcontracted.
- What constitutes an unacceptable or incomplete report.

A useful assignment can disconfirm a target or explain uncertainty. Do not reward a consultancy simply for producing a longer list.