How to test ads before spending

An ad test usually means paying to run several versions on the ad platform and seeing which one gets clicks or sales. That’s the real test, but every version you run costs money. Testing before you spend, also called ad pre-testing, means asking people which versions they prefer, so you only pay to run the strongest few.

Two kinds of ad test

A pre-test asks people. You show your ad versions to a group and they choose. It’s quick and costs little, and it tells you what people prefer when they look on purpose.

A live test pays for real impressions. The ad platform shows each version to strangers as they scroll, and you compare clicks, sign-ups or sales. It measures behaviour, which is what you care about in the end, but it takes budget and time.

The two work best together. Use the pre-test to cut eight or twelve versions down to two or three, then run those live. Preference testing vs A/B testing compares the two kinds of test beyond ads.

Step 1: make versions you can compare

Start from one idea and change one thing at a time: the headline, the image, the first second of the video, or the offer. If every version differs in everything, a winner tells you nothing you can reuse.

Keep the format the same across versions. Compare square ads with square ads and 15-second videos with 15-second videos.

Step 2: ask the right people

Ask people who look like your buyers: customers, newsletter readers, followers, members of a community you run. Your team and friends know the product and know you, so they judge differently. If you want their view, run a separate poll.

If you don’t have an audience yet, you need strangers from a target group, and that means a paid panel. Services such as PickFu sell those answers. Pairpoll doesn’t.

Step 3: ask a buyer’s question

The question shapes the answer. “Which ad is best?” invites people to judge design. “Which ad would make you stop scrolling?” or “Which would make you click?” is closer to what the ad has to do.

Step 4: show ads the way people meet them

In a feed, nobody studies an ad. For video ads, let each version play once with no replay or pause. In Pairpoll that’s the “each video plays once” setting. For image ads, a quick pick between two is closer to scrolling than a form that asks people to score each one.

Step 5: read the result honestly

  • Drop the versions nobody picked. That’s the most reliable thing a small poll tells you.
  • If one version leads by a wide margin, fund it.
  • If two are close, treat them as a tie and run both live. How many voters do you need? shows how easily a narrow lead from a few people comes from chance.

Step 6: run the finalists live

Run the two or three versions that survived on the ad platform, with the same audience and budget for each. Compare them on the result you care about, such as cost per sign-up or per sale. That’s your answer. If the live winner surprises you, the pre-test still saved money: you didn’t pay to run the versions that lost early.

What a pre-test can’t tell you

People asked to judge look harder than people scrolling past. People who know you react differently from cold audiences. And saying you’d click is not the same as clicking. A pre-test reduces the guesswork before you spend. It doesn’t replace the live test.

Running the pre-test in Pairpoll

  1. Upload 2 to 16 ad versions, as images or as MP4 or WebM videos.
  2. Write the question, choose a top 1, 3 or 5, and for videos turn on “each video plays once”.
  3. Share the link. Voters type a name and start, with no sign-up. They pick the better of two until their top list is settled.
  4. Read the overall ranking and each voter’s top list, then fund the leaders.

Ad creatives shows a sample poll you can try. For video ads, see video ad testing.

All guides

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.