How FixEvidence ranks the next action
Safety exclusions run first. Applicable actions are then evaluated using model and symptom relevance, official support, already-attempted checks, smoothed outcome evidence, durability, cost, difficulty and risk. Confidence is calculated separately from the recommendation.
Recommendation and confidence are different
A high expected success estimate can still have limited confidence when the sample is small or the match is broad. FixEvidence never hides this distinction.
Outcome denominator
Each repair attempt is stored independently. A complete journey can include several failed attempts, a final resolution and 30/90-day durability follow-ups.
Evidence signals
Email, model photo, error photo, receipt, invoice, technician and follow-up verification are stored separately. Verification strengthens confidence but never erases the raw attempt count.