AI Follow-up Questions: What Triggers One and What Doesn't¶

AI Follow-up Questions (see AI Follow-up Questions: What They Are and Where They Work for the basics) don't fire on every open-ended answer by default - a follow-up appears when a condition you've set is actually met, which means you have real control over how often respondents encounter one.

The Trigger Condition¶

When you enable AI Follow-up on an open-ended question, you configure a condition based on the respondent's answer - checking, for example, whether the response meets a particular criterion you define. Only an answer that meets your configured condition triggers a follow-up; an answer that doesn't meet it moves the respondent straight to the next question, same as if the feature weren't enabled at all.

Custom AI Instructions¶

Beyond the trigger condition, you can give the feature custom instructions describing what kind of follow-up you want - steering it toward probing pain points, asking for a specific example, clarifying an ambiguous answer, or whatever direction is most useful for your research question. This is what keeps the generated follow-up relevant to your specific goal rather than generic.

The Max Follow-up Limit¶

You can also cap how many AI-generated follow-ups a single respondent can receive across the survey. This keeps a respondent who gives several trigger-worthy answers in a row from being probed repeatedly - past the cap, the survey simply moves on without generating another follow-up, regardless of whether later answers would otherwise have qualified.

Setting Up a Trigger Condition Well¶

A condition that's too broad means most respondents get a follow-up, which can make the survey feel longer than intended. A condition that's too narrow means the feature rarely fires at all, and you're paying for a capability you're barely using. Previewing your survey and testing with a range of realistic answers - short, vague, detailed - is the most reliable way to check that your condition is catching the answers you actually want probed.

FAQ¶

Can I set a different trigger condition for different open-ended questions in the same survey?
Yes - the condition is configured per question, so different open-ended questions can have entirely different triggers.

What happens if a respondent's answer is right on the edge of my condition?
This depends on exactly how you've defined the condition - test with borderline example answers during preview to see how your specific setup handles them.

Does the Max Follow-up limit apply per question or across the whole survey?
It's a cap on the respondent's total follow-ups across the survey, not a separate allowance per question - check your specific survey's configuration for how it's applied.

If a follow-up doesn't trigger, does the respondent see any indication that it could have?
No - a respondent whose answer doesn't meet the trigger condition simply moves to the next question, with nothing suggesting a follow-up was possible.

Does testing trigger conditions in preview mode cost credits?
Check your survey's credit estimate before testing - since AI Follow-up generation is a credit-based action (see What Actually Consumes Your Credits (and What Doesn't)), triggering it during preview may behave differently from a live respondent session, so review your specific setup before assuming preview runs are free.

For the full picture of setup and where this feature applies, see AI Follow-up Questions: What They Are and Where They Work.