Skip to content

Preview and results

When a participant takes a study, FromUsers roughly applies rules in this order:

  1. Variables resolved — randomisation, URL, cookies, sample, formulas
  2. Section display logic — hide whole sections that do not apply
  3. Section randomisation — shuffle or pick pages/subsections within visible sections
  4. Page display logic — hide or skip pages (including managed study-group rules)
  5. Page qualification / screener — pass or fail before main content
  6. Page block randomisation — shuffle or pick blocks on the current page
  7. Question display logic — show or hide individual questions
  8. Compare assets — resolve prototype for the visible page
  9. Branching — after answers, jump to another page or section

Most studies use one or two layers. Use this order when debugging preview behaviour.


Live studyBuilder preview
Group assignmentServer permuted-block balancing via assignment indexClient-side simulation
PersistenceStable across reloads (same session)Per-tab; Restart preview re-draws random assignments
ResultsGroups stored on responsesPreview responses may be stored separately

Testing tips:

  • Turn on Follow logic in the preview panel to simulate participant paths
  • Use study group selectors to jump between arms
  • Click Restart preview to simulate a new participant assignment
  • Test every arm before publishing

SetupWhat appears in results
No randomisationStandard response data
Shuffle / subset (no groups)Responses without a named experiment arm
Randomisation groupsGroup variable values on each response — filter and compare by group
Compare assetsPrototype shown per participant
Display logic onlyNo extra randomisation field unless driven by a variable you defined

For export and dashboards, use randomisation groups or an explicit randomisation variable whenever you need to segment by condition.


  • Do not mix direct pages and subsections in one randomised section — restructure first
  • Thank you page — keep outside randomised pools
  • Name variables clearly (task_variant not v1) when using groups across compare assets and sections
  • Test every arm before publishing — especially when combining section randomisation, compare assets, and display logic
  • Screener first — place qualification pages before randomised main content so disqualified participants do not consume assignment slots
  • Preview is not identical to live balancing — use live pilot responses to verify group splits at small sample sizes