1. Pick who sees it
Create a variant and send it to a share of your visitors, such as half of them. Each visitor keeps seeing the same variant on every visit.
Survey A/B testing
Test two ways of asking, and keep the one that works
Run survey A/B testing on the wording, the length, the layout or the look. Usertick shows each variant to the visitors you choose, compares the answers question by question and tells you when a difference is clear.
One survey, more than one version
Variants live inside the survey you already have. Each one changes only what you edit in it, and everyone a variant doesn’t match sees Base.
From a variant to a verdict
Make a variant, change one thing, and read the result once enough answers are in.
2. Change one thing
Edit the variant like any survey. Every field you changed is marked Different in this variant, and Reset brings back the Base value.
3. Let the answers come in
Usertick waits until each side has 30 answers before it compares anything, and shows how far each variant still has to go.
4. Read the verdict
Once a difference is clear at 95% confidence, Results says which version scores better, and on which question.
What to test in a survey
A variant can change almost everything people see. These tests answer the questions teams ask most often.
The wording of a question
Compare a question that leads people with a neutral one, or a long wording with a short one, and see whether the scores move with the words.
How long the survey is
Drop the questions you can live without in a variant, here on the survey page a link opens. With an even split, compare how many responses each side collects and how long people take.
Step by step or Chat
Ask the same questions as a conversation in a variant, and compare the answers and the time people spend on them.
The layout of the survey page
On a link survey, try another page layout, another accent or your logo in a variant, and compare the responses and the answers each page collects.
The look of the widget
Try another theme, a dark look, a header bar or another corner of the page, and compare the responses and the answers each look collects.
The title and the buttons
Change the survey’s title, its intro line and its button texts in a variant, and see which invitation gets more responses.
Read the comparison question by question
The Variants tab in Results compares every group with everyone else and sums up the result in one sentence.
The right test for each question
Averages, NPS, answer shares and single options each get the statistical test that suits them, with a correction when many options are compared at once.
Better or Worse where it counts
Where a higher score is better, such as NPS or ratings, the result reads Better or Worse. On other questions it says whether the answers differ.
Nothing called too early
A result needs at least 30 answers on each side and 95% confidence. Until then, each group shows how many answers it still needs.
On your dashboards too
Add the Variants card to a dashboard to keep a running test in front of your team, with each group’s score and verdict.
Show each group its own survey
Variants also tailor a survey to a country, a device, a page or a user trait, with no second survey to keep in sync.
A shorter survey on phones
Send phone visitors a variant with fewer questions while desktop keeps the full survey. A fix to Base still reaches both.
An extra question for Pro users
Match the plan your app passes to Usertick, and ask Pro users about a feature only they have. Their question appears only in their variant.
Act on the result
Usertick never moves traffic on its own. You decide, and the next step is one action away.
Make the winner your Base
Apply to Base moves the winning variant’s changes and questions into the Base survey and removes the variant, so everyone gets the version that works.
AI
Hear when there’s a winner
The AI reading points out a variant that scores clearly higher or lower, and suggests applying it to Base or deleting it.
AI
Test a rewrite before you replace it
When AI suggests a clearer wording for a question, Test as variant shows it to half of your visitors before you replace the original.
How to run a survey A/B test you can trust
- 01
Change one thing at a time
Test the wording, the length or the look, never all three at once, so you know which change made the difference.
- 02
Split your visitors evenly
Give the variant half of your visitors, so both sides collect answers at the same pace and their numbers compare.
- 03
Wait for the verdict
A result needs 30 answers on each side and 95% confidence. Stopping after one good day reads luck as a win.
- 04
Run it for at least a week
Visitors can answer differently on a weekday and on the weekend, so keep the test running across every day of the week.
- 05
Keep the answer options the same
Change the words of a question but keep its options, so the answers stay comparable. A variant with other options can’t be compared directly.
- 06
Apply the winner, then test again
Make the winner your Base, delete what lost and start the next test from there.
Frequently asked questions
Add a variant to a survey, change what you want to test in it and choose who sees it, such as half of your visitors. Each visitor sees the first variant whose conditions match, and everyone else sees Base. Results then compares the answers of every group, question by question.
Up to eight, plus the Base survey. Variants are matched in order, so a visitor who fits two of them sees the first.
Once both groups have at least 30 answers and the difference is clear at 95% confidence. Until then, the group shows Collecting with the answers it has so far. Usertick never shifts traffic on its own, so the decision stays with you, and Apply to Base makes the winner your new Base.
Usertick counts responses, not views, so it doesn’t show a response rate per variant. With an even split, the number of responses each side collects, the median time and the answers to each question show which version works better.
No. Triggers, targeting, frequency and availability are shared by every variant, so a test compares the survey itself.
Yes. A share of visitors is stable for each visitor, so the same person keeps seeing the same variant on every visit and the comparison stays clean.
Link surveys show variants, including the survey page’s layout and logo. In an email survey, the first question inside the email always shows Base. Page and user trait conditions match only on your website.
Creating and publishing variants is on the Business plan, and every plan can read variant results.
Listen better, act faster, and build what truly matters.
Start free, see what your customers mean, and get each answer to the person who can act on it.
- 150 responses a month on Free
- Every channel and AI analysis
- No card needed