"Tested both networks for two weeks, picked the winner"

"Tested both networks for two weeks, picked the winner"

Granted: an A/B test beats vibes, and two weeks of data feels like enough to call it.

It usually isn't — ad CPMs swing on a roughly monthly and quarterly cycle. A two-week window can land entirely inside one network's strong half of the month and the other's weak half.

🚩 Hidden variable: the calendar. End-of-quarter brand budgets, holiday demand spikes, and mid-month pacing resets move CPMs 20-30% with zero relation to which network you ran. Your "winner" might just be whoever you tested over the 15th.

The honest comparison runs both across the same full month, ideally two — or at minimum aligns the test windows day-for-day.

Not saying A/B testing is wrong — saying a two-week window compares calendars as much as networks.