The method is selected to answer the client’s decision question. Models are not treated as evidence on their own: inputs, assumptions, calibration limits, uncertainty and authorised use are stated explicitly.

1. Define the decision

Agree the technical question, the decision it supports and what would constitute a useful answer.

2. Review the available data

Confirm sources, units, time coverage, quality, compatibility, gaps and permitted use.

3. Identify supported explanations

Compare candidate mechanisms against observed behaviour and established engineering principles.

4. Select the appropriate analysis

Use only the calculations or modelling required to distinguish the relevant explanations.

5. Test assumptions and uncertainty

Check sensitivity, calibration evidence, alternative interpretations and limits of transfer.

6. Deliver findings and next actions

Provide a concise technical record, review discussion and practical next step.

Quality, traceability and data handling

Versioned inputs and calculation basis. Clear distinction between measured, interpreted and modelled information. Documented assumptions and exclusions. Independent checks proportionate to scope. Corrections and revisions recorded before final issue.

Client data should not be sent through the public enquiry form. Confidential transfer, permitted use, retention and return or deletion arrangements are agreed before data is received.