Optimizing AI

Optional retest · video · Illustrative study · synthetic data

Compare two video cuts before you launch.

The suggestion was “show the product earlier.” Retest the late-reveal cut against an earlier reveal and compare the same moment.

Illustrative product and creative research preview

Video edit

Illustrative study · synthetic data

COMPARING When to reveal the product

DAYLIGHT / CONCEPT FILM60 SEC
A better start.
DAYLIGHT08:00 Your morning, on time.

Meet Daylight.

City morning
Rushed commute
0:42 · Reveal
Original fictional storyboardCUT A
Illustrative viewer artwork; not a study participant or signal readingIllustrative viewer artwork

0:42 · Product reveal

Paused for 4 seconds at the reveal

0:42
0:000:200:401:00
Example data · fictional

“I wasn’t sure what this was for until the end. Then I went back to check what the device was.”

Synthetic participant feedback · V3

What to test next Researcher interpretation

Ask other participants whether the opening felt slow before the product appeared.

A study and report concept, not a verified production workflow. Optional pulse is secondary context, not evidence of an emotion. Confirm demo availability in the call.

The full video comparison

This fictional 60-second spot illustrates a study and report concept. Confirm demo availability and the exact workflow for your project in the call.

Illustrative study · synthetic data

Late reveal — the product first appears at 0:42

  1. 0:00

    City morning

  2. 0:10

    Missed alarm

  3. 0:21

    Rushed commute

  4. 0:32

    Late to meeting

  5. 0:42

    Product reveal

  6. 0:52

    End card

bpm707478revealV30:000:150:300:421:00
Cut A: Late reveal — the product first appears at 0:42. Example data · fictional. Participant V3 playback: 0:00 Started playback; 0:27 Scrubbed forward 6 s; 0:42 Paused for 4 s at the reveal; 1:00 Watched to the end card. Optional pulse estimate from 72 to 76 bpm at most. Signal quality: Partial, 0:14–0:19 excluded (participant turned away). Illustrative study · synthetic data.
Optional pulse estimate, V3 Product reveal Example event 0:42Signal quality: Partial — 0:14–0:19 excluded (participant turned away)

Observed · participant V3

  • Scrubbed forward once during the commute scenes.
  • Paused for about four seconds when the product appeared at 0:42.
  • Optional pulse estimate rose around 0:42 and settled by the end card.

“I wasn’t sure what this was for until the end. Then I went back to check what the device was.”

Interpretation · researcher note

The scrub and the comment suggest the opening ran long before the product gave it context. The pulse change at 0:42 shows only that something changed near the reveal — not what the viewer felt.

Next step: Ask other participants whether the opening felt slow before the product appeared.

Example values for Cut A
Playback events and pulse estimate for Cut A. Illustrative study · synthetic data.
TimePlayback (V3)Pulse est.
0:00Started playback72 bpm
0:27Scrubbed forward 6 s71 bpm
0:42Paused for 4 s at the reveal75.5 bpm
1:00Watched to the end card73 bpm

More on what this preview does with data: Privacy and sample data.

Text version

The same example, in words

Both cuts are 60 seconds long and share one time scale. 0:42 is the marked example event in each. Illustrative study · synthetic data.

Cut A: Late reveal

Late reveal — the product first appears at 0:42.

Observed

  • Scrubbed forward once during the commute scenes.
  • Paused for about four seconds when the product appeared at 0:42.
  • Optional pulse estimate rose around 0:42 and settled by the end card.
  • Participant comment: “I wasn’t sure what this was for until the end. Then I went back to check what the device was.”

Interpretation

The scrub and the comment suggest the opening ran long before the product gave it context. The pulse change at 0:42 shows only that something changed near the reveal — not what the viewer felt.

Cut B: Earlier reveal

Earlier reveal — the product first appears at 0:12.

Observed

  • No scrubbing during the first 40 seconds.
  • Rewound five seconds at 0:42, the “on time” scene.
  • Optional pulse estimate changed slightly at 0:12 and again near 0:42.
  • Participant comment: “I got what it was early. I rewound the end bit because I missed how it actually got me there on time.”

Interpretation

Seeing the product early appears to have given the story a frame. The rewind at 0:42 points to a payoff scene that may be too fast — worth testing a slower version.

How a team could use it

A video comparison in four steps

  1. 01

    Pick two cuts

    For example a late versus an earlier product reveal, with everything else kept the same.

  2. 02

    Show each cut to its own group

    Participants watch at their own pace and answer a few questions afterwards.

  3. 03

    Line up the evidence

    Playback actions, comments and — only for participants who opt in — a pulse estimate, on one shared time scale.

  4. 04

    Decide what to test next

    The comparison points to moments worth a closer look. It doesn’t tell you why any one viewer did what they did.

What this can’t tell you

A pause, a rewind or a pulse change shows that something happened at a moment. It doesn’t show what a viewer felt or why they acted — that comes from what they tell you. The pulse shown here is an optional webcam-derived estimate for research context: it is not a medical measurement, and it is sensitive to lighting and movement. In a real study, every participant must be able to decline the optional signal and still take part; no production consent workflow is verified or claimed on this site. See the methodology and limitations, or the onboarding sample study.

Designed for voluntary product and creative research. Optimizing AI is not offered for monitoring employees or students or making decisions about individuals.

Book a demo

A conversation with the founder, booked through Calendly.