When Something Goes Wrong

Most studies run cleanly, but here's what the exceptions mean and what to do.

A survey failed

If the survey couldn't complete, the study shows a survey failed state with a short reason. Common causes are temporary — a service was briefly unavailable, or a stimulus image couldn't be read. What to do:

• If it points to a stimulus problem, check the image is still present and re-add it, then start again.
• Otherwise it's usually transient — start the study again after a short wait.

An analysis failed

If the survey finished but the analysis stopped, the study shows analysis failed with a Resume button. Resuming picks the analysis back up rather than re-running the whole survey, so it doesn't cost you the survey again. Click Resume and let it continue.

An abandoned study

An abandoned study is one that was stopped and won't continue. It stays in your library as a record. Start a fresh study if you still need the answers.

A design version conflict

If two people edit the same study, or you edit it in two places, the design can get out of sync. The editor shows a conflict with three choices:

• Keep mine — use your version.
• Load current — take the other saved version.
• View differences — see exactly what changed (fields, and questions added, removed, changed, or moved) before deciding.

If you try to leave with unsaved changes, you'll be prompted so nothing is lost. To avoid conflicts, agree who's driving a study design at any one time.

Good to know

• A failed survey and a failed analysis are different: only an analysis failure offers Resume. A survey failure means starting the survey again.
• When in doubt about a message, Errors & Edge Cases explains the user-facing ones in plain language.