Variation sets allow you to run the same test on different versions of a design, or multiple tests on the same design and ensure that all participants across all tests in the set are unique. This ensures that participants have not seen the design before, so will not be biased by previous tests.

This is great for iterating on a design or testing variations to see which performs best.

Click Make a variation from your preferred test. If you're on the dashboard, click Create variation set.

Give your variation set a name.

Click Copy variation and create variations.

Click Save changes on the modal.

Similar to other tests, a variation set has a unique link you can share with your audience to direct them to take a test from your set. Toggle on and copy this link from the Recruit participants tab.

Recruits who visit this link will be taken to one random variation. We'll split them evenly across variations. Your recruited participants will only be allowed to participate in one of the tests in this set, whether they revisit the set's recruitment link or visit the recruitment link of any test in the set.

If you are on the Pro plan, you can request demographic information from your participants by turning on Request demographics from the Recruit participants tab of your set. Information collected will be available with results: age, country, gender, and education level.

Unlike recruiting from your own audience, when you order responses from the UsabilityHub panel you cannot place an order against the variation set to receive an equal split of unique participants across your tests. Instead, you will need to place individual response orders for each test within your variation set.

For example, if you want to survey a total of 100 unique participants across 4 tests equally in a variation set, you will need to place an order for 25 responses on each test. By placing your tests inside a variation set we ensure that each response comes from a unique panelist.

Please note that this restriction can affect order delivery times if you order many responses (or from niche demographics) for tests within a variation set. The estimated turnaround time for the order you place on your last test may be longer than your first test as those panelists have been removed from the pool of eligible participants.