> ## Documentation Index
> Fetch the complete documentation index at: https://docs.getversive.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Logic & branching

> Build conditional paths, screen participants out, and pipe answers forward

Question logic lets each question branch based on the participant's answer. Rules are evaluated **top to bottom — the first matching rule wins**. If no rule matches, the interview simply continues to the next question.

## Conditions

The conditions available depend on the question type:

| Question type                                   | Available conditions                                                |
| ----------------------------------------------- | ------------------------------------------------------------------- |
| Multiple Choice                                 | Always · Equals · Includes (multi-select) · Fewer than N selections |
| Rating Scale, NPS, Star Rating, Number          | Always · `=` · `>` · `>=` · `<` · `<=`                              |
| AI Question, Exploratory, Open Text, Short Text | Always · **AI rule**                                                |
| All other types                                 | Always                                                              |

### AI rules (smart conditions)

For open-ended questions, you can write a condition in plain language — for example, *"If the participant mentions price as a concern"* — and Versive's AI evaluates it against the response. AI rules are evaluated in order along with your other rules.

<Tip>
  Keep AI rules specific and observable ("mentions a competitor by name") rather than interpretive ("seems unhappy"). Specific rules evaluate more reliably.
</Tip>

## Actions

When a condition matches, one of five actions runs:

| Action              | What happens                                                                                                                 |
| ------------------- | ---------------------------------------------------------------------------------------------------------------------------- |
| **Go to question**  | Jump ahead to a later question, skipping everything in between.                                                              |
| **Move to group**   | Jump ahead to a question group — the interview enters at the group's first question and continues from there.                |
| **End interview**   | Finish the interview immediately and show the thank-you message.                                                             |
| **Screen out**      | End the interview and show the screen-out message — the response is marked screened-out and doesn't count toward your quota. |
| **Redirect to URL** | Send the participant to an external URL (not available on AI Question or Exploratory questions).                             |

### Where jumps can land

Jumps are **forward-only**: a rule can only target a question or group that comes after the question it belongs to, so the last question of a study has no jump targets. When your study uses question groups, two more rules apply:

* **Randomized groups** — questions inside a group with randomized order can't be targeted directly, since their position differs per participant. Target the group itself instead: each participant enters at whichever question their shuffled order puts first, so randomization is preserved.
* **Empty groups** can't be targeted — there's no question to jump to.

The target pickers only offer valid destinations, so you don't need to memorize these rules. If a later edit invalidates a saved rule — say, deleting the target or enabling randomized order on its group — the logic editor flags the rule with a warning.

## Screening participants

To build a screener, put your qualifying questions first and add **Screen out** rules to each. Screened-out participants see the customizable [screen-out message](/studies/settings#study-messages), and their responses are excluded from completions and insights by default.

## Reusing earlier answers (mentions)

To pipe an earlier answer into a later question, use **mentions**: type `@` in the question editor and pick a previous question to insert a reference to its answer — or to a specific option or property of it. Each participant sees their own earlier answer woven into the question text ("Why did you choose the *Pro* plan?").

Answers from these types can be referenced: Open Text, Short Text, Multiple Choice, Matrix Rating Scale, Card Sort, Ranking, and Allocation.

## Good to know

* The logic editor warns you about rules that can no longer fire: targets that were deleted, moved to before the question, or moved inside a randomized group, plus rules on the last question.
* Logic is language-aware — rules are evaluated against the participant's study language.
* Chaining multiple rules on one question is fully supported; order them from most to least specific since the first match wins.