Among 1,962 exposed users on a post-registration offer page, visible video controls recorded a 1.75% purchase rate versus 1.42% without the control bar; the historical display marked the difference not significant. We kept the video fixed and changed navigation. Keep the usable player while testing a specific restriction separately; video length is a different comparison.
Why we tested restricted playback
The video pitched an offer on a post-registration thank-you page. From earlier work, we expected keeping viewers in the intended sequence might help them hear the pitch before deciding. We tested that expectation here rather than assuming it would transfer.
Same video, different player controls
| Arm | Player experience |
|---|---|
| Control | Default player with visible control bar |
| Treatment | Control bar removed to prevent skipping or fast-forwarding |
We followed experiment exposure through checkout to purchase success, using completed purchases after the pitch as the primary outcome.
Visible controls led; the difference was not significant
The report summary covers nine days and 1,945 participants, with purchase rates of 1.76% for control and 1.43% without controls. The saved public result display contains 1,962 exposed users and these results; its observation window and settings have not been reconciled with the report.
| Arm | Exposed users | Checkout views | Purchase rate | Relative change | Historical 95% relative-lift interval | Win probability |
|---|---|---|---|---|---|---|
| Visible controls | 973 | 57 | 1.75% | Baseline | Baseline | 71.93% |
| No control bar | 989 | 50 | 1.42% | −18.98% | −51.33% to +35.15% | 28.07% |
Only 31 purchases occurred across both arms. The saved display gave control a 71.93% historical probability of being best and marked the difference not significant.
The historical relative-lift interval includes a possible improvement for the restricted player. Keep the usable player while testing a specific restriction, rather than treating this as proof that removing controls always hurts purchases.
Keep the player usable while testing the pitch
Our explanation was that losing navigation could frustrate viewers or make it harder to revisit a claim. We did not measure trust or attention directly. A person watching longer because seeking is blocked is not necessarily more persuaded.
Keep play/pause, volume, captions, seeking and full-screen controls available by default. Check keyboard operation, visible focus, control labels and zoomed layouts; offer playback speed where the player supports it. Accessible media players need controls people can use without a mouse and with assistive technology. (+)
Test a specific restriction, not an invisible player
If there is a concrete reason to test seeking, change that setting separately from hiding the entire control bar. Keep the video, player styling, page copy, offer and CTA timing fixed. Verify the setting on mobile, desktop and keyboard before sending traffic.
Track starts, progress, seeking, completion, CTA exposure, exits and playback errors alongside purchases. Check device and connection differences so a player failure does not look like a lack of interest.
Use the PostHog experiment guide to run the comparison, then the VSL-length page if the next question is what to cut from the pitch.