Ad creative testing: see who stops, and why, before you spend

Most teams learn that an ad does not work by paying for impressions. The platform shows the click-through rate a few days later, but not why people scrolled past, and a small budget split across several versions rarely reaches a clear answer. A check before the spend will not replace that live test, but it catches the ads that were never going to work: the ones people cannot read in a second, do not believe, or cannot connect to a product.

• Image ads, social posts, and videos of up to 3 minutes.
• See who would click, stop to engage, glance, or scroll past, and why.
• For a video, see when each person would have scrolled away.
• Run the next version on the same simulated people and see who changed their mind.

The costliest ad problems show up in the first second. People cannot tell what is being sold, the text on the image is too small to read on a phone, the offer is buried, or the ad looks like a dozen others and gets scrolled past without a thought. A crowd of skeptical simulated people is good at catching these, because each of them has a reason to keep scrolling.

They also catch mismatches between the ad and the audience: a promise that excites a founder and bores the finance lead who signs off, or a term one group reads easily and another does not. The report breaks reactions out by role, age, gender, and country, so a split shows up instead of averaging away.

It cannot tell you which of two close ads will win. In a pre-registered test on real Wikimedia fundraising banners, Mimiq's forecast picked the banner that raised more on 54.1% of 37 tests, not clearly better than a coin (run C4 in the public register). Pre-registered runs on emails and text messages found it no better than chance at ranking small wording changes, and on real web page tests it did not beat the simplest guess.

It cannot give you a click-through rate, a cost per click, or a return on ad spend. The counts in a report describe 25 simulated people, not your campaign. Only real impressions measure those.

Its forecast has held up on one kind of test so far. On 1,000 real headline A/B tests held out from its development, Mimiq's forecast picked the headline more readers clicked 76% of the time; the best rule of thumb, picking the headline written first, got 61%. Headlines are not ads, so treat a call between two ads as a first read, and the objections as the useful part.

Screen before you split the budget. If you have six ideas, test them all and drop the ones that confuse people or that nobody believes. Put the live budget behind the two or three that survive, where real clicks decide.

Fix before you scale. When a live ad underperforms, a test tells you what people misread or doubted, the part the platform numbers leave out. Fix that one thing, check it on the same simulated people, then relaunch.

A video test takes a file of up to 3 minutes. Your browser takes up to ten frames from it, more of them in the first seconds, where most viewers decide, and the file itself never leaves your device. Each person says when they would have scrolled away and what they were thinking at that moment.

Put the product and the offer in the first frames. If most people have scrolled away before the part you are proudest of, the report will show it.

Your first test is free, no account needed. A free account adds 100 people. After that, an ad test on 25 people uses 25 credits, and packs start at $29 for 500 credits. Credits never expire, and there are no seats or subscriptions.