A/B Tests (Random Assignment)

ARTICLE

A/B Tests (Random Assignment)

A/B Tests

Paid feature
Log in to see if this feature is included in your plan.

Our A/B Test feature allows you to randomly display different versions of a question, image, or free-standing text, and set the percentage of respondents that will be shown each variable. When you analyze your results, you can assess how the different stimuli affected responses.

Creating an A/B Test

Set up your A/B Test on the Design Survey page.

  • Image A/B Test
  • Text A/B Test
  • Question Text A/B Test
  • Video A/B Test

As you add variables, the respondent percentages will split evenly across all variables. However, you can enter your own percentages if you prefer. Variable percentages must total 100%. These percentages determine the probability of a new respondent seeing that variable when they take your survey.

Previewing A/B Tests

Before you send out your survey, preview your survey design a few times to see what your survey will look like to respondents. You'll be shown different A/B Test variables each time you preview your survey.

As people take your survey, SurveyMonkey randomly assigns each respondent to be shown a certain variable based on the percentages you define.

Analyzing A/B Test Results

There are several areas in the Analyze Results section where you can view and analyze the results of your A/B test.

  • Under the Question Summaries tab, you can see the number and percentage of respondents that viewed each variable.
  • In the CURRENT VIEW section of the left sidebar, you can Filter by A/B Test or Compare by A/B Test.
  • Under the Individual Responses tab, you can see which variable each respondent was shown.
  • You can filter or compare by A/B test for these question types: Multiple Choice, Checkboxes, Star Rating, Dropdown, Matrix/Rating Scale, Ranking, and Image Choice.

You may notice that the distribution of respondents that saw each variable doesn't exactly match the percentages you defined. The more responses you collect, the more likely it is that your results will reflect the percentages. Since variables are randomly assigned, we do not guarantee the split will be exact, particularly for smaller numbers of participants. We suggest collecting at least 50 responses.

TIP! See if there are significant differences in how respondents who saw different variables answered your survey.

Exports

When a Filter by A/B Test rule is applied, you can only export summary data in PDF or PPT format. However, the A/B Test questions themselves are excluded from PPT exports since there are no charts and data tables to display in the presentation.

When a Compare by A/B Test rule is applied, you can only export summary data in PDF format.