# Assessment Centre Exercise Source Separation

Keep behavioural ratings from different exercises traceable instead of flattening them into one unexplained score.

## Scope
Preparation and review only. No account action or professional decision.

## Inputs
Candidate/test IDs, exercise types, observed behaviours, selector IDs, job criteria and approved combination policy.

## Workflow
1. Separate each exercise and evidence source.
2. Link observations to job criteria before interpreting ratings.
3. Keep an overall score held when no approved combination policy is supplied.
4. Return a source matrix for experienced selectors, not a hiring recommendation.

## Output contract
candidate | exercise | observer | behaviour | criterion | rating_source | combination_hold

## Fictional acceptance test
Fictional candidate has group-task rating 4 and individual-task rating 2. Without a defined combination policy, do not invent an overall 3. Preserve both observations and their different exercises.

## Review gates
Keep original values, source IDs and version dates. Unknown facts stay held. Use qualified review for hiring, legal and accounting conclusions. No send, share, sign-in, scheduling, ledger change or payment. Fictional contract review is not a live account compatibility test.

## Method source
https://www.cipd.org/en/knowledge/factsheets/selection-factsheet/

Checked October 12, 2026. Oct12 live AIS Search Console: marketing Copilot37 impressions, Planable/Claude38, Planable/ChatGPT24, campaignbrief12, investoragent19, MuseIndia100. Social calendar/reporting topics follow parent-query leads; other professions follow documented workflow needs, not measured exact intent. Site impressions are not monthly volumes; exact topic volume is unmeasured.