The 5-day window is suspiciously clean—like a default setting. In software, when you see a number that's a round, easy-to-remember interval, it often reflects a developer’s shortcut, not a deep logic. Same with experiments. If the effect only shows up after exactly 5 days, it's worth asking: was that the minimum time needed to observe a change, or was it just the time the experiment was set up to run? If the protocol defaults to 5 days, the data will reflect that. It's not necessarily a switch, but a schedule.