A writer is in a cycle using AI-Assisted Reviews, but their review answers are blank or no draft appears. A blank answer does not always mean something is broken. Several of the causes below are expected behavior.
Access and Availability
- Review writers can check their own drafting choice and question types.
- Checking the cycle's AI settings and drafting audience requires permission to manage that review cycle, including review admins and collaborators where applicable.
- Available in the Total Platform and Perform & Goals pricing packages.
- Applies to manual review cycles. Lifecycle Reviews are excluded from AI-Assisted Reviews and are unaffected by these settings.
Diagnose
Check: The writer chose to write the review themselves
- How to check: Ask the writer whether they selected Write it myself instead of Start drafting when they opened the review.
- If true: This is expected. Answers begin blank by design. The writer can still generate individual answers later, use the review assistant, open the Context library, add context, and draft again.
- If not: Continue to the next check.
Check: The review type does not receive AI drafting
- How to check: Confirm which review the writer is completing.
- If true: Peer reviews and upward reviews use the new review experience without AI drafting, Moments, or the review assistant. This is expected. AI drafting is available for self reviews, manager reviews, and additional-manager reviews.
- If not: Continue to the next check.
Check: AI drafting is not enabled for this writer
- How to check: In the cycle's settings, confirm AI drafting is enabled, then confirm the writer is included in the drafting audience.
- If true: A cycle can use AI-Assisted Reviews while excluding certain writers from AI drafting. Use the fix below.
- If not: Continue to the next check.
Check: The question is not an open-text question
- How to check: Look at the question type in the review template.
- If true: This is expected. 15Five AI drafts open-text answers only. Rating-scale questions are intentionally not answered by AI, and every rating remains the writer's own selection. If the template contains only ratings or scores, there is nothing for AI-Assisted Reviews to draft.
- If not: Continue to the next check.
Check: There is not enough usable context for that question
- How to check: Open the Context library and look at what is available for the question. Some questions have much richer context than others.
- If true: A newer employee, someone whose work happens mainly in a system that is not connected, or a question that is poorly represented in the available evidence may not have enough information for a useful draft. 15Five AI may leave that answer blank rather than filling the gap with something plausible. Use the fix below.
- If not: See If It Still Fails below.
Fix
Fix for: AI drafting is not enabled for this writer
- Open the review cycle's settings.
- Confirm AI drafting is enabled for the cycle.
- Check the drafting audience and include the writer, either by adding them directly or by adding a group they belong to.
- Save your changes. For an active cycle, see Edit settings in an active performance review cycle.
Fix for: there is not enough usable context
- Add a Note with the information that should inform the answer.
- Add a File to the Context library.
- Review the available 15Five Data.
- Include another relevant Moment.
- Ask 15Five AI to draft the answer again, or write the answer yourself.
Validate
The writer opens the review and a draft appears for the eligible open-text questions. Where the cause was the drafting audience, the writer's name is included in the cycle's drafting audience after saving.
If It Still Fails
Contact the 15Five Support Team with the review cycle name, the participant, the review writer and which review they are completing, which questions have no draft, and what was expected. Drafting that fails repeatedly across several people or questions, with no obvious explanation in the available context, is worth reporting rather than retrying.
Not Covered Here
This article does not cover cycles that are showing the legacy review experience, or drafts that appear but are inaccurate or weak. See Troubleshoot: A review cycle is using the legacy review experience.