Ad & Video Research

Test whether your video earns the watch, holds attention through the moments that matter, and actually lands the brand and the message — measured second-by-second, not after the fact.

What we test

Direct viewing analysis

The spot (or one of several cuts) is shown directly to participants. We get the scene-by-scene attention curve, the engagement trace, where attention drops off, whether the brand and benefit moments land, and post-viewing recall and comprehension.

In-feed / scroll behavior

Your ad placed inside a simulated social feed. Participants scroll naturally. We measure whether they stop, how long they watch before swiping, and what about the opening frames earned (or lost) the watch — the headline question for social-media advertising.

Applies to TV and broadcast ads, social-media video, product demos and explainers, and brand films. Either setup can be run as A/B variant testing across multiple cuts.

Typical questions we answer

These are representative examples, not a fixed set. If your question is not listed, it can most likely still be answered.

"Where in our ad does attention drop?"

Who we help: Creative directors and brand teams

Second-by-second attention curve overlaid on the spot, with the precise scenes where viewers disengage identified — so the re-edit conversation starts from data, not intuition.

"Is the brand actually seen at the brand moment?"

Who we help: Brand directors and brand-tracking leads

Frame-level attention to logo, brand mark, and product at the exact seconds of the brand reveal — separating "the spot was watched" from "the brand was registered."

"Which cut of the ad performs better?"

Who we help: Brand managers and agency teams running A/B cuts

Side-by-side attention curves, brand-moment fixation, recall, and stated preference across cuts, with statistical-significance tests on the differences that matter.

"Does our social ad stop the scroll in the first 1.5 seconds?"

Who we help: Social media and performance teams

Stop rate, watch duration before swipe, and what part of the opening frames earned the watch — tested inside a simulated feed, not as a forced view.

"Are viewers emotionally engaged in the moments that matter?"

Who we help: Brand planners and insights leads

Arousal trace (rPPG) and facial-expression dynamics aligned to the spot timeline, identifying peaks at the benefit moment and dips at the dead air.

"Did viewers actually understand the message after watching?"

Who we help: Brand and product marketing teams

Aided and unaided recall and message take-out, segmented by whether viewers actually fixated the key copy or visual — distinguishing "did not see it" from "saw it, did not get it."

"Why is the spot scoring well in surveys but underperforming on air?"

Who we help: Insights and media planning teams

Reconstructs the viewing moment to separate "liked but not branded" from "branded but not understood" from "understood but not memorable" — three different fixes.

"Does the ad work across age groups and audience segments?"

Who we help: Brand and segmentation leads

Attention, engagement, recall, and preference broken down by demographics or self-reported familiarity, surfacing where the spot lands for one segment and misses another.

Anatomy of a typical study

Let's take a common ad question: "Is our finished cut delivering attention to the brand and benefit at the right moments, and does the message land?" Here is how an ad pre-test answers that.

Sample & setup

We recruit 120–180 viewers from your target audience to watch the spot in their own browser, alongside a category-control ad to give a benchmark for normal attention behaviour. Eye tracking runs continuously across every frame, and we can optionally layer in arousal measurement (rPPG pulse) and facial-expression tracking. A short post-view survey closes the session — aided and unaided recall, message take-out, and impression.

What we measure

Frame-level attention — where on screen viewers look at every moment
Attention curve across the timeline, with peaks and drops identified scene-by-scene
AOI dwell on brand mark, product, benefit copy, and call-to-action during their on-screen windows
Optional arousal trace (rPPG) and facial-expression dynamics aligned to the timeline
Post-view aided and unaided recall, message take-out, comprehension, and stated preference
Whether the per-cut differences hold across audience segments (demographics, prior brand familiarity)

What you get

We deliver a written findings report, starting with an Executive Summary & Recommendations section, followed by our in-depth analysis. This includes the frame-level attention curve overlaid on the spot timeline with brand, benefit, and CTA windows annotated, AOI metrics during each window (time-to-first-fixation, dwell, recognition latency), optional arousal and expression traces, recall and comprehension results, per-segment breakdowns, and statistical-significance tests. Heatmaps and scan-paths are added for key frames. Both raw and processed data are appended to the report (gaze, fixations, arousal, expression, and survey responses). We present the findings to your team in a 60-minute live readout, and answer any follow-up questions for 2 weeks after.

Typical outcome

Our goal is to always provide you with the most actionable insights possible. For ad pre-tests this often means it is one of three possible outcomes:

The cut lands as intended: attention is on the brand and benefit at the right windows, post-view recall and message take-out are strong, and the result holds across segments. The action is to commit media spend.
Brand-attention gap: viewers watch through and rate the spot well, but attention is somewhere other than the brand or benefit at their moment, or recall is weak — the spot entertained more than it informed. The cut-level recommendations are specific: move the brand reveal to a high-attention moment, repeat the benefit visually rather than only in voiceover, tighten a specific dip. Re-test the new edit before commit.
Creative isn't carrying: attention drops early and never recovers, recall is weak across segments, and the message does not survive the view. The lever is the creative direction itself, not an edit — recommendations are about reworking the cut or testing a different concept altogether.

Typical engagement

Timeline

4–6 weeks

Investment

€15–30k

Typical sample

120–200 participants

Single-cut pre-tests are usually medium complexity. In-feed scroll studies and multi-cut A/B programmes add complexity on the build and analysis side. Cost depends mainly on that complexity (number of cuts, optional biometric layers, segment depth) and the sample size the question requires.

See the full engagement process on the homepage.

Have an ad or video research question?

Let's discuss it.

Submit a project enquiry →