When to test an ad

The earlier you test, the cheaper the fix. Most teams test at one of four points.

Before the shootAnimatic or storyboard
Check the idea and the story arc land before you spend on production.
Before the media buyFinished cut
Confirm the ad works, find the weak seconds and fix the edit while you still can.
Choosing between editsTwo or three versions
Each viewer sees one version, so you can compare them fairly and pick on evidence.
Making cut-downs30s to 20s, 10s or 6s
The second-by-second curve shows which moments carry the ad and which can go.

Ways to test an ad, compared

Ad testing has split into approaches that answer different questions. Here is where facial coding with real viewers fits.

Comparison of ad testing approaches
ApproachReal viewers?What it measuresExamples
Predictive AINoA predicted score from a model trained on past ads. Results in minutes, useful for screening many ideas.iMotions Predictive AI, System1 Test Your Ad Screen, Kantar LINK AI
Survey with self-reported emotionYesWhat viewers say they felt, for example by choosing a face or word after watching.System1 FaceTrace, standard ad surveys
Managed facial codingYesFacial expressions coded while viewers watch, usually reported as proprietary scores.Adverteyes (formerly Realeyes PreView), Element Human
Facial coding softwareYour own sampleExpression analysis you run and interpret in-house.iMotions with AFFDEX
EchoDepthYes44 FACS Action Units per frame, a second-by-second Emotional Response Curve with the method published, and the Cavefish Emotional Index from what viewers say.Run as a study for you

Descriptions of other providers are taken from their own websites in October 2026 and summarise their main public offer. Realeyes moved its ad testing business into a separate company, Adverteyes, in November 2025. Affectiva's ad testing is now offered through iMotions.

Why teams choose EchoDepth

Three things you won't get from a predicted score.

Measured on people

Every result comes from viewers who actually watched your ad, matched to your audience and screened for quality.

Open to inspection

Every score breaks down to the seconds and Action Units behind it. How the response curve is built is published, so your analysts can check our work.

Consent first

Viewers opt in to each study. No video of faces is stored, and we never code employees or students.

How a study runs

Remote, browser-based and managed for you from brief to debrief.

  1. Brief and sample plan

    Tell us the ad, the audience and the decision. We propose a sample, usually 15 or more viewers per audience segment.

  2. Recruit and screen

    Testers from our panel and partner panels, matched on age, gender, region and category use.

  3. Viewers watch at home

    Consent, a webcam setup check, a short neutral baseline, then your ad and a short survey.

  4. Readout and debrief

    The Emotional Response Curve, the Cavefish Emotional Index, peaks, drops, brand moment and stated vs felt, walked through on a call.

Ad testing questions

More on the measurement itself is in the methodology, and on costs in pricing and plans.

Before the money that's hardest to recover is spent. Test an animatic before the shoot if the idea is new, and test the finished cut before the media buy. Testing two or three edits side by side at either stage is the most common brief.

Yes. Vertical ads are shown to viewers at their native aspect ratio, and the second-by-second curve is especially useful for the first two or three seconds, where short-form ads win or lose attention.

Yes. Animatics produce a usable emotional curve, and testing at this stage lets you fix the story before production. Expect lower overall intensity than a finished film, so compare animatics with animatics.

In-platform A/B tests compare live ads on clicks and conversions after you have paid for the media. Pre-testing shows the ad to a matched sample before launch and explains why it works, moment by moment, so you can fix it before spending.

EchoDepth recommends 15 or more viewers per audience segment for each ad or edit. Every group you want to report on separately, such as under-35s and over-35s, is its own segment and needs its own 15. We agree the sample with you based on your audience and the decision you need to make.

A report with the second-by-second Emotional Response Curve (valence, arousal and dominance), the peak and drop moments with the Action Units behind them, the brand moment, the Cavefish Emotional Index from what viewers say afterwards set against what their faces did, and a debrief call.