Beside the answer being rated. For a shared form, include that answer’s context.

Check reported failures against the original answer before adding them to an evaluation set. These self-selected ratings cannot establish model accuracy or grant permission to reuse conversations for training.