Free Preference Test Tool

A preference test shows people two to six versions of a design and asks which one they prefer for a specific reason, such as which makes it clearer how to book. It is a quick way to settle a design argument with evidence from the people who will use the thing.

Yes. Building a test and previewing it as a participant costs nothing, and there is no credit card step. A study takes up to 1,000 responses. You sign in with an email link only to get the share link, and your results are not locked. There is no participant panel, so you send the link to your own testers.

Upload two to six designs, write one specific question, and preview it. Then sign in, copy the share link and send it to participants. Open the results to see votes per design, the reasons people gave and the significance hint.

One specific, answerable question beats "which do you prefer?". Try "which makes it clearer how to book?" or "which would you trust with your payment details?". The optional follow-up (why did you choose it?) is where the useful detail comes from, so read the reasons, not just the percentages.

Screenshots or mockups of the same screen at the same size, so the only difference is the thing you are testing. Upload PNG, JPG or WebP (they are resized for you), or paste a link to an image that is already hosted. Participants can tap any design to see it full size, and a phone frame suits mobile designs.

The tool will not call a winner under 20 responses, because a handful of votes can swing either way by chance. With 20 or more it shows a p-value hint: an exact binomial test for two designs and a chi-square test for three to six. Treat it as a hint, not proof, because your participants are self-selected.

Each participant sees the designs in a random order, and the tool records the order they saw, so position bias (people favouring the first or left-hand option) is spread across the designs instead of favouring one. Results are always counted against the original A, B and C.

Yes. The show-one-each option gives every participant a single design instead of a choice, with the least-shown design served first. Results then compare the groups rather than call a winner, and the tool wants at least 10 responses per design before it compares them.

Votes per design, the reasons people gave and the significance hint. Responses export as CSV.

It tells you which design people pick and what they say about it, not whether the design works. Pair it with a first-click or task-based test before you ship, and remember there is no participant panel here, so you recruit your own.

Paid tools mostly add a participant panel, team workspaces and support. This one doesn't recruit for you. The designs, randomised order, votes, reasons, significance hint and CSV export are free.