Home › How scoring works

How scoring works

Every score on this site is an AI evaluation out of 100 unless it is explicitly labelled as a community result. We never present an AI estimate as a user rating.

The scores

AI Quality - an AI evaluator's judgement of structure, clarity, specificity and reusability.
AI Usefulness estimate - the evaluator's estimate of how useful the output is for its stated purpose.
Evaluation Confidence - how confident the evaluation is, given prompt completeness and testing.
Goal Match - computed per search: how well a prompt fits the goal you typed.
Community results - real "worked / didn't work" votes from visitors. Until a prompt has votes it shows "No votes yet", with the sample size shown once votes exist.

Why AI estimates?

New prompts have no usage history. An AI evaluation gives an honest first signal, and community votes refine it over time using a weighting that prevents a prompt with two votes from outranking one with five hundred.

Limitations

AI scores are estimates and can be wrong. Always review a prompt's output before relying on it, especially for financial, legal or real-estate use.