Upload your ad. Read what it does.
Soma runs your video through a brain-response encoder and hands back the readout: a lane heatmap of the response, the weak windows where it sags, and edit candidates that were tried and re-measured on your file. Measured readouts, not promises.
Three readouts, straight off your file.
- what
- Which processing lanes light up, second by second, across the whole ad.
- why
- You see where the ad is working the brain — and where nothing much is happening.
- what
- The stretches where the measured response sags below the rest of the ad.
- why
- A retention graph shows where people leave. The readout shows what went quiet first.
- what
- Edits the pipeline actually tried on your file, each one re-scored after the change.
- why
- Every candidate carries its before-and-after readout, so you judge the measurement, not our taste.
A readout is a measurement, not a promise.
The audit shows what the model measured on your ad, and where the measurement is thin it says so. We do not promise that acting on it will improve your campaign, and we do not dress a model output up as a guarantee. When an edit candidate moves the readout, you see the numbers it moved — that is the whole claim.
If that honesty is what you want from an ad tool, the audit is free and takes one upload.
One upload, one readout.
Sign in, upload the ad from your dashboard, and the pipeline runs the encoder over it — the same read it runs on every ad it serves. The readout lands in your library, pinned to your timeline second by second. How the read itself works is on the science page.