Test your ad creatives before you pay to run them

Put up to 16 versions of an ad in one poll and send the link to customers, followers or your team. Each person picks the better of two until their top list is settled, and you get each voter's list and the overall ranking.

Voters don't sign up. Files are deleted after 30 days.

Which ad would make you stop scrolling?

Pick 1 of 4

Sample pictures. Nothing you pick is saved.

When this helps

You have more ad versions than budget. A designer sent eight, or an image tool gave you twelve, and you can afford to run two or three. Picking by gut means your favourite gets the money. Asking a few people who look like your buyers costs nothing, and it can show that the version you were about to fund is one nobody else picks.

How a poll for ad creatives works

  1. Export each version as an image, or as an MP4 or WebM video. One poll holds images or videos, not both.
  2. Create a poll and write the question voters will see, for example “Which ad would make you stop scrolling?”
  3. For video ads, turn on “each video plays once”. Both ads in a pair then play through once, with no replay or pause. Nobody studies an ad frame by frame in a feed either.
  4. Send the link to people who match your buyers: a customer email list, a community you run, or followers. Voters type a name and start. They don’t sign up and they never see your file names.
  5. Open the results. Fund the two or three versions at the top, and drop the ones that were never picked.

What a poll tells you, and what it doesn’t

A poll shows which ad people prefer when they look at it on purpose. A live ad test shows which ad strangers click while they scroll past. Those are different questions, and a poll doesn’t replace the live test. Use the poll to cut a long list down. Then spend your test budget on the versions that survived, with real traffic on the ad platform.

Your own audience also knows you already. People who have never heard of you may react differently. How to test ads before spending covers both steps in more detail.

Tips

  • Change one thing between versions when you can: the headline, the image or the offer. Then the result tells you which change worked.
  • Ask the question a buyer would ask. “Which would make you stop scrolling?” gets a different answer from “Which ad is best?”
  • Your team knows the product too well to judge an ad like a stranger. Ask them in a separate poll if you want their view too.
  • For short video ads, a top 1 or top 3 keeps the poll quick. Each pick means watching two videos.

Common questions

Is this an A/B test?

No. An A/B test runs ads on live traffic and measures clicks or sales. Pairpoll asks people which version they prefer. It’s a quick check before you spend, so the live test starts with stronger options.

How many people should I ask?

A handful of people is an opinion check, not proof. If two versions end up close, treat them as a tie. How many voters do you need? explains how to read small numbers.

Can people outside my audience vote?

Anyone with the link can vote, and you set the maximum number of voters. Pairpoll has no paid panel of respondents. If you need answers from strangers who match a target group, a paid panel such as PickFu fits better.

Can an AI agent set up the poll?

Yes. An agent can create the poll, open it and read the results through the API, the CLI or the MCP server. See the agent docs.

Other things to rank

Make your own poll

Free. Upload 2 to 16 images or videos and share one link. Voters don't sign up.

Create a poll

Pairpoll uses Google Analytics cookies to see which pages people visit, only if you accept. Details are on the privacy page.