Split is a block that lets you create several variations inside and split the flow of respondents between them. Each respondent randomly ends up in one of the variations and only sees its content. A handy way to run A/B testing within a single study.
What the split block is used for
Split lets you compare several versions — of mockups, prototypes, copy, videos — without creating separate tests for each variation. Typical scenario: one group of respondents sees one design variation, another sees a different one, and both groups answer the same follow-up questions. All in one study.
How the split block works
A split can have two or more variations. You can add any blocks inside each variation: questions, prototypes, context, choice, and so on.
When a respondent reaches the split block, we automatically route them to one of the variations. The distribution is even — each variation gets roughly the same share of respondents. After completing their variation, the respondent continues the test from where the split block ends.
Heads up: each respondent sees only one variation. If you've allowed multiple responses and a respondent has already taken the test once, the assigned variation won't change — the same person will see the same variation.
How to create a split block
Open your test in the editor.
Click the add block button and pick Split under the "Other" section.
Two variations are created by default. If you need more, check the section below.
Managing variations
Add a variation — click "Add variation" under the last one.
Duplicate a variation — hover over the variation's header and click the copy icon. All blocks inside the variation will be copied.
Delete a variation — hover over the variation's header and click the delete icon. You can only delete a variation if there's more than one in the split block.
Logic and the split block
After a split block, you can set up logic — rules that define which block a respondent goes to next — based on which variation they ended up in. Two conditions are available:
"variation is" — the respondent ended up in the specified variation
"variation is not" — the respondent ended up in any other variation
This lets you ask different follow-up questions after the split block depending on which variation the respondent saw.
For example: you're comparing two landing page variations. After the split block, you want to ask respondents whether the product description was clear — but the wording on the landings is different.
With logic, you can ask different follow-up questions: one for respondents who ended up in variation A, another for those in variation B. And common questions (like demographics) can stay the same for everyone.
What you'll see in the report
Results for the blocks inside each variation are shown separately in the report. That way you can compare responses across variations and see how the differences affected the results.
Split block vs split group: what is the difference
Both tools deal with splitting respondents, but they solve different problems.
Split block splits the flow of respondents within a single test. Good for comparing a piece of the research — a prototype, a wording, a video — while the rest of the questions stay the same for everyone.
Split group combines several separate tests and guarantees that respondents don't overlap between them. Good when you need separate tests with separate reports — or when you have one audience and multiple studies to run.
| Split block | Split group |
What it compares | A part of one test | Whole separate tests |
Shared part of the test | Yes (before and after the split) | None — tests are independent |
Report | One shared, variations side by side | Separate report for each test |
No respondent overlap | Within a single test | Between separate tests |
When to go with a split block
For example: you're testing a new onboarding and want to see which of two intro screens works better. The rest of the test — instructions, tasks, final questions — is the same for everyone. A split block fits best here: two onboarding variations inside one test, and you set up the shared questions just once.
When to go with a split group
For example: you're comparing two completely different website prototypes. Their use cases are different, tasks are different, questions are different too. It's easier to make two separate tests in a split group and invite respondents with a single link — each person will end up in just one test.
Another example: you've got one audience channel (say, a chat) and five different studies you need to run in parallel. Instead of five separate messages — create a split group and share one link. We'll evenly distribute the respondents, so you won't burn out your audience with repeated requests.
More about split groups — in the Split groups article.