Troubleshoot Survey Flow

Flow issues usually come from missing destination blocks, variables that appear too late, condition arms that are not connected, or quota paths that do not end where expected. Troubleshoot from the respondent path first, then inspect the Flow canvas with the exact answers that produced the issue.

Start in Preview, not in the diagram. Once you can reproduce the path, open Flow and check the block, condition set, connector, quota, and end screen that should have handled that respondent.

• Respondents skip a block they should see: the connector may point around the block, or a branch condition may be using the wrong option label or variable.
• Everyone lands in the same arm: the condition set may be set to by-variable assignment, a random percentage may be missing, or the fallback path may be catching all respondents.
• A screener does not screen anyone out: the eligibility question may appear after the Flow split, or the branch may be checking the display label instead of the stored value.
• Quota-full respondents continue anyway: the quota rule may be attached to the wrong condition arm, or the quota-full connector may be missing.
• Preview differs from live collection: changes may be saved in the editor but not pushed to the published version. Review Publish and Share a Research Survey.

1. Step 1: Reproduce the path in Preview. Use Preview and write down the answers that lead to the issue.

   Step 1: Reproduce the path in Preview

   Block Builder

   Flow

   Preview

   Deploy

   Export

   Preview

   Walk through respondent-facing behavior in the same preview surface.

2. Step 2: Check the source variable. Confirm the screener, answer, quota status, or custom variable appears before the Flow split and has a clear question name.

   Step 2: Check the source variable name

   Block Builder: Voter Registration Screener

   The registered_voter variable drives the Flow branch — eligible respondents continue, ineligible route to a screen-out end screen.

   Multiple Choice

   Required

   Page Break

   Question text

   Multi-select

   Options

   Add QuestionPull Questions

   Closing text

   Example only.

   Redirect (optional)

3. Step 3: Inspect the condition set. Open Create Experimental Conditions and confirm assignment mode, arms, percentages, and fallback behavior.

   Step 3: Inspect condition set in Flow

   Block Builder

   Flow

   Preview

   Deploy

   Export

   Flow Canvas

   Interactive routing surface from the app, running in no-persist help mode.

   Consent and Study Intro

   1 question

   Political Knowledge Block

   2 questions

   Control Message

   1 question

   Treatment Message

   1 question

   Outcome Measures

   2 questions

   Survey Ends

   Anchors explicit end of survey

4. Step 4: Check every connector. Make sure each arm routes to the correct block, including treatment, control, ineligible, quota-full, and end screen paths.

   Step 4: Check all arm connectors

   Block Builder

   Flow

   Preview

   Deploy

   Export

   Flow Canvas

   Interactive routing surface from the app, running in no-persist help mode.

   Consent and Study Intro

   1 question

   Political Knowledge Block

   2 questions

   Control Message

   1 question

   Treatment Message

   1 question

   Outcome Measures

   2 questions

   Survey Ends

   Anchors explicit end of survey

5. Step 5: Review quotas separately. If the issue happens only when a cell is full, check Set Survey Quotas.

   Step 5: Review quota configuration

   Block Builder

   Flow

   Preview

   Deploy

   Export

   Deploy

   Configure publishing, links, quotas, and theme from the Deploy tab.

6. Step 6: Submit test responses. Send one test response through each path, then use Export to confirm condition assignments and route-relevant fields are recorded.

   Step 6: Check condition fields in Export

   Block Builder

   Flow

   Preview

   Deploy

   Export

   Export

   Download analysis-ready response data from the Export tab.

• Variable appears after the split: move the question earlier or route from a later block.
• Arm reaches the wrong material: reconnect the condition arm to the intended treatment or end screen.
• Quota path is confusing: create a dedicated quota-full end screen.
• Random assignment looks uneven in a small pilot: review Sample Allocation Across Condition Arms before changing the design.

• Branching and Skip Logic for Research Surveys
• Online Survey Experiments
• Survey Data Quality Checks