How FixEvidence ranks the next action

Decision order

1. Apply safety stops.
2. Validate appliance, brand and exact-model context.
3. Recognize supported error codes without treating a code as proof of one failed part.
4. Remove actions already attempted.
5. Apply symptom, prerequisite, model-family and source-reviewed rules.
6. Add the closest available moderated outcome layer: exact model, model family, brand, then general guidance.
7. Subtract cost, difficulty and risk penalties.

Recommendation and confidence are different

A high-ranked action can still have limited outcome confidence. Match confidence explains how specifically the technical rule fits; outcome confidence explains the amount and quality of community evidence.

Minimum denominators

FixEvidence does not display a success or recurrence percentage from fewer than five matched observations. Raw counts remain visible. Higher confidence requires larger samples, verification signals and, where available, exact-model relevance.

Repair journeys

Attempts are stored separately, including failures. A final resolution and follow-up never erase earlier failed attempts; otherwise success rates would be biased upward.

Last updated: August 28, 2026.