Every research method is wrong in its own particular way. Card sorts show how people group things when grouping is the task, which is not how they behave when they are trying to get something done. Analytics show what people did, not what they wanted. Interviews record what people say they do. A finding that survives across several methods is worth acting on; a finding from a single method is worth investigating. This prompt sorts findings into that distinction, converging, diverging, complementary, and attaches a confidence level to each.
Feed it findings rather than raw data from four sources at once, and state plainly which decisions the synthesis will inform, since that is what determines whether a given disagreement matters. Read the diverging-evidence section first and trust it least: the model will offer explanations for the contradictions, and those explanations are hypotheses, not findings. A real contradiction between methods usually means the methods measured different things, and working out which one answers your question is a judgement that stays with you.
Confidence comes from methods that fail differently agreeing anyway. One method agreeing with itself is just one method.
Fill in your own details below; the prompt updates as you type. When it's ready, copy it into Claude or whatever AI tool you use.
Combines findings across research methods, separating high-confidence convergences from single-method signals.