# Evidence matching for Profit Model diagnosis (Core-light v0.4.4)

Use the index to shortlist at most a few plausible Profit Models from the business problem and reported facts. Read their cards before making a model claim.

1. Match an actual business mechanism (what changes, for whom, and what economic effect is plausible), never a matching word, product feature, sector, or company size alone.
2. Treat synonyms and user-specific descriptions as possible evidence only if the card's required discriminator is satisfied. Do not translate terms into canonical graph variables or universal node/edge types.
3. Record separately: user-reported facts, assumptions, candidate explanations, missing evidence, contradictory evidence, and planned interventions. Time order alone does not establish causation.
4. Prefer the distinguishing question from the strongest confusable alternative. Ask no more than one high-value question in a turn, and provide a useful interim explanation.
5. Apply the six-dimension readiness gate and contradiction checks in diagnostic-rules.md before naming a diagnosis. Allow no-fit/indeterminate; no match is a valid result.
6. Where two models appear relevant, report only mechanisms supported independently by case evidence; do not infer model composition, compatibility, or new models from catalog relationships.
7. Do not claim to simulate business dynamics, discover a novel Profit Model, or optimize interventions. Such capabilities are outside this standalone diagnostic Skill.
