Glossary
User Feedback
Explicit or implicit signals from users that indicate whether they found a model response helpful, harmful, or otherwise notable.
User feedback encompasses all the ways users signal their reaction to model outputs — clicking thumbs up or down, flagging a response, writing a complaint, or abandoning a conversation. It’s a noisy but valuable signal: users know what their actual experience was, even if they can’t always articulate why a response was good or bad. Interpreting user feedback requires care because it’s often biased toward extreme reactions (very good or very bad), reflects individual preferences that may not generalize, and can be missing for the vast majority of interactions. For behavior architects, user feedback is most valuable when combined with other signals — log analysis, evaluation data, and expert review — rather than treated as a standalone source of truth.