Skip to main content
Good diagnostics triangulate evidence. No single interview, process document or dataset should automatically be treated as the whole truth.

​
Unstructured evidence

Use stakeholder interviews to understand:
  • decisions;
  • exceptions;
  • workarounds;
  • handoffs;
  • pain points;
  • unofficial tools;
  • causes that are not visible in system logs;
  • differences between intended and actual work.
Consent-based screen observation can add granular task evidence where appropriate, especially when people struggle to explain repetitive or detailed work verbally.

​
Structured evidence

Operational data can help quantify:
  • volumes;
  • timestamps;
  • cycle times;
  • variants;
  • rework;
  • status changes;
  • handoff patterns;
  • exceptions.
Useful sources can include CRM, ERP, ticketing, workflow and transactional systems.

​
Do not collect data for its own sake

Every requested field should help answer the agreed business question. Large, poorly scoped data requests slow engagements and create unnecessary security and cleaning work.

​
Evidence quality checklist

For each important finding, ask:
  • What is the source?
  • Is it direct evidence or interpretation?
  • Is it representative or exceptional?
  • Does another source support or contradict it?
  • What would change our conclusion?

​
Consultant behaviour

When two stakeholders disagree, do not choose the account you prefer. Treat the disagreement itself as evidence and investigate why the process differs.
Specific Iris setup steps, participant controls and supported structured-data formats should be documented in Iris Academy and marked [PRODUCT DETAIL TO CONFIRM] where the current product behaviour is not established.