NBCUniversal · Peacock ads and playback
Testing new ad breaks with a real control group.
An A/B test of HLS interstitial ads on Peacock VOD. I co-designed the test and its cohorts with data scientists, then drove the fixes that came out of it.
HLS interstitialsA/B testingData scienceAd playback
48MPeacock paid subscribers
Comcast, Q2 2026. Product scale, not my users
The problem
Peacock was moving ad breaks on VOD to HLS interstitials. I wanted to know, with a real control group, whether that changed how people binge.
What I owned
I co-designed the test and its cohorts with data scientists. I worked with them to implement, measure and monitor it.
How it went
Test against a control group
I ran it as an A/B with a control group, so the binge result could be read cleanly.
Choose the cohorts with data science
I helped pick the cohorts up front instead of leaving that to the analysis step.
Act on what the test shows
I drove the fixes before the rollout, instead of treating the test as only a readout.