When evaluating performance, it is useful to measure not only how quickly a draft is produced, but also how long it takes the assignee to edit and approve it. If requests are misclassified and have to be reassigned, or people need to recheck missing information, faster generation does not necessarily translate into less time spent on the overall task. During an initial rollout, teams can select one type of recurring request and track how often information is missing, edits are needed, or reviews must be repeated.