THE SHORT ANSWER

Start with the observed symptom, locate when and where it occurs, identify what changed, generate several possible causes, note constraints and choose a discriminating next test. Keep observation separate from explanation.

Move from symptom to next test

Use SYMPTOM → CONTEXT → CHANGE → POSSIBLE CAUSES → CONSTRAINT → NEXT TEST. At each step, ask what is observed and what is inferred.

Diagnostic moves across contexts
ContextSymptom questionDiscriminating follow-up
MarketingWhere did conversion change?Did traffic mix, page behavior or measurement change first?
WorkflowWhere does work wait or return?What input or decision is missing there?
TeamWhich commitments are being missed?Is the pattern tied to clarity, capacity, dependency or capability?
ProductWhich users fail at which step?Can we reproduce it under the same conditions?

Keep several explanations alive

‘Why?’ repeated mechanically can produce a tidy story rather than a reliable diagnosis. Complex outcomes can have interacting causes. Ask what else could explain the pattern and what evidence would distinguish the alternatives.

Evidence & context: National Academies of Sciences, Engineering, and Medicine

Choose information that can change the diagnosis

A useful next test is not merely more data. It separates plausible explanations. Compare affected and unaffected cases, inspect the point where behavior diverges, or change one controllable condition while watching an agreed measure.

Public appraisal guidance emphasizes objectives, options, evidence, risk and uncertainty. The same discipline helps prevent a diagnosis from becoming a preferred solution in disguise.

Evidence & context: HM Treasury and Government Finance Function

Build a one-page diagnostic

  1. Write one observable symptom without a cause word.
  2. Add where, when, for whom and how often it appears.
  3. List three plausible explanations, including a measurement problem.
  4. Name the constraint that limits investigation.
  5. Choose the smallest test that could separate two explanations.

For a workflow-specific method, continue to How to Find Workflow Bottlenecks.

Sources & further reading

  1. The Green Book 2026

    HM Treasury and Government Finance Function. Current official guidance on objectives, options, evidence, risk, uncertainty, appraisal and evaluation. It is designed for public decisions, while the general reasoning principles are adapted here.

  2. Reproducibility and Replicability in Science

    National Academies of Sciences, Engineering, and Medicine. A consensus report on evidence, uncertainty, transparency, reproducibility and replication. Scientific standards need proportionate adaptation outside research settings.

Examples and exercises are illustrative unless attributed to a source. No independent expert review is claimed.

A correction, a counterexample or an experience worth sharing?

Join the conversation ↗