Reading Your Results
Refreshing and exporting a report after cleanup
Three sub-pages, three caches
The exportable report is not one single calculation. It is built from three separate sub-pages, Search Intel, Segmentation, and Overview, each of which processes the underlying wave data in its own way and caches its own results. That caching is what makes the pages fast to load, but it also means a change you make elsewhere, like cleaning up entities, does not automatically ripple through to all three at once.
This caching applies specifically to the exportable report's three sub-pages. The analysis tab's own co-occurrence and direct attribution view isn't part of it — it queries the underlying entity data directly, or automatically re-aggregates in the background right after a cleanup action, so a cleanup pass typically shows up there without waiting on the refresh sequence below. That sequence only matters once you're ready to generate the exportable document.
When this becomes relevant
This becomes relevant any time you have gone back and adjusted something after a wave has already run, most commonly after an entity cleanup pass where you moved unrelated entities to "other" or merged duplicate variants together. That cleanup changes the underlying data, but each sub-page only picks up the change once you tell it to refresh.
Refresh in the right order
Refreshing the sub-pages out of order, or only refreshing one of them, can leave your export in an inconsistent state where some sections reflect the cleanup and others still show the old numbers. To avoid that, refresh in a fixed order:
- Search Intel first.
- Segmentation next.
- Overview last.
Each of these can depend on data computed by the one before it, so working through them in this order means every sub-page is refreshed against inputs that are already up to date.
Once all three have been refreshed, in that order, only then export the report. Exporting before the refresh sequence is complete will bake the old, pre-cleanup numbers into whatever document you generate, and you would need to reopen the process and export again to fix it.
Refreshing analysis is different from running another wave
This sequence updates analysis and reports from responses you already collected. Run another wave when you need a fresh set of model answers, for example after a launch or a change in the news. A later wave in the same study reuses its study slot, while its new responses use your organisation's shared funds. Review the wave estimate before running it. AI analysis actions can also use funds; refreshing a report is not a promise of free AI processing.
Keep a study when you need its history for later comparison. Archiving it changes how you organise your work, but does not release its study slot. Check Billing for your organisation's current study capacity.