Three different states
They look similar and are not:
- Generating — the interview finished and the evaluation is running. A few minutes.
- Not started — no evaluation was requested. Usually an interview that ended without recorded answers.
- Failed — it was attempted and did not complete.
The report tab says which. A single spinner for all three would be the actual bug.
Generating for a long time
Give it several minutes; a long interview with many answers takes longer than a short one. If it is still going after roughly fifteen, treat it as failed.
Failed
Retry from the report tab. Generation is idempotent — retrying re-runs the evaluation over the stored answers and does not re-interview anyone or charge again.
If it fails repeatedly, open a ticket with the interview link. A repeated failure is usually something specific about that sitting rather than a general outage, and it is diagnosable from the record.
For candidates: no report at all
Whether you see one is the organisation's setting, made per stage. Some release automatically, some release when they choose, and some never do. A missing report is more often that setting than a failure.
Your own practice interviews are different — those are yours, and the report appears when it is ready.
The report is there but a number looks wrong
Open the per-question breakdown. Every score carries its reasoning, and the reasoning usually explains the number — either the answer missed something, or you and the evaluation are weighting the role differently. Those need different responses, which is why the reasoning is shown rather than only the score.