Guide
UX researchers have run five-second tests for years: flash a design, then ask what stuck. Every feed runs a harsher version of that test on every ad, all day — except nobody gets asked anything, and failure just looks like a scroll. Here's how to run the test on purpose, before the feed runs it on your budget.
The panel version — Lyssna (formerly UsabilityHub) made it famous — shows a respondent your design for about five seconds, hides it, and asks questions: What was this for? What do you remember? Whose brand was it? It measures recall and comprehension of a glimpse, and for static work — landing heroes, banner layouts, packaging — that's exactly the right question, answered in human words.
An ad in a feed doesn't get a guaranteed five seconds — it gets a fraction of one, and it has to earn the rest. The failure mode isn't "they saw it and misunderstood"; it's "they never stopped," or they stopped and left at second four. A recall questionnaire can't see either event, because the leave decision happens mid-play, wordlessly. For video, the five-second test becomes a curve question: how hard does the opening grab, and where exactly does the grip fail?
Best when the deliverable is comprehension in customer language: show the opening frame or a static variant, ask what people took away, read the answers. Priced per response, results within hours. Use it on your first frame, your thumbnail, or a static ad where "did they get it?" is the whole question. Its limit is the one above: paid attention isn't feed attention, and a static glimpse can't tell you where a 30-second cut bleeds viewers.
PreTestAds runs the behavioral version: AdCortex™, trained on fMRI brain-response data, scores any finished ad — filmed, designed, or AI-generated — and returns Hook Strength (your first three seconds, the model's five-second test), the full attention curve with the exact drop-off second, and a percentile against 76 top-performing TikTok ads. Minutes per test, about $2.45 each, up to 8 free credits. It won't tell you what viewers understood — that stays the panel's job — it tells you whether they'd still be watching when your message lands.
For creative that matters, run both halves in order. First the behavior gate: score the cut, fix the hook if the first seconds are weak — you can even test hook copy before filming. Then the comprehension check: five-second-test the opening frame with humans and confirm the message that survives the glimpse is the one you meant. Both together still cost a rounding error next to the $500+-per-variant live test — which is the version of this test you run by accident when you skip the first two.
Upload your ad and see your Hook Strength and drop-off second in minutes — up to 8 free credits, no credit card.
Test Your Ad Freeor read a live sample report first — no signup
A first-impression research method: show someone a design for about five seconds, take it away, and ask what they remember and understood. Tools like Lyssna (formerly UsabilityHub) run it with human panels. It's a staple for landing pages and design comprehension.
Partially. It tests recall and comprehension of a glimpse, which suits static designs. Video ads fail differently — viewers scroll away mid-play — so the question isn't only 'what did you remember?' but 'at which second did you leave?'. That's an attention-curve question, which panels answer awkwardly and attention-prediction models answer directly.
Two routes. Panel route: a Lyssna-style test shows your static or opening frame to real people and collects what they recall — great for comprehension. Prediction route: upload the finished ad to PreTestAds and get Hook Strength (the first 3 seconds), the full attention curve, and the exact drop-off second in minutes, from about $2.45 per test with up to 8 free credits.