Dayparting with Smart Bidding: keep ad schedules or drop them?

I'd stop hand-tuning hour-level bid adjustments and use ad schedules for one reason only: whether someone can actually answer the phone. Hourly adjustments mattered back when manual CPC was the default. With tCPA or Maximize Conversions, the algorithm already sees conversion rate by hour and prices each auction on it, so a 20% bump at 9pm mostly adds noise. In a small account you can easily burn weeks reading a pattern that isn't really there. The exception is lead gen, where a missed call at 11pm means a lost lead by morning. In that case a schedule that pauses the campaign outside office hours is a business rule, not an optimization. Has anyone seen hourly adjustments still beat the algorithm on volume?

Agreed on the phone case, but a schedule has a cost the bidder can't see. The hours you switch off produce no conversion data, so when you reopen them Smart Bidding is still working from daytime results. If the phone rule is real, keep the schedule, and expect a relearn after any schedule change.

Theo, I don't think the relearn is much of a problem. Smart Bidding doesn't wipe what it learned about a paused hour. When you reopen it, the conversion history from before the schedule is still there, so the real downside is that it may price 11pm on older estimates. For a phone-driven business that's a small cost, since the schedule was there to stop those leads from mattering anyway.

The bigger risk is the reverse. If the hours you paused were already weak, the missing data costs you almost nothing. But a schedule you didn't strictly need can quietly cut volume during hours that were fine. So before you blame the bidder for slow performance after a schedule change, check whether the schedule removed traffic you actually wanted.

Agreed on the stale-estimates point. For phone businesses, the hour the bidder learns from may not be the hour the phone rang. A plumber's ad clicked at 9pm often turns into a call at 9:40, or a booking the next morning. If you import call or offline conversions, check whether the timestamp is the click time or the conversion time, because that decides which bucket gets the credit.

If the credit lands on the click, a schedule that pauses 9pm can also cut the hour that quietly brings in next-day jobs, and your daytime numbers will look worse than they are. So before judging a schedule change, pull a few weeks of call conversions and line them up against click hours. It takes about twenty minutes and tells you whether the bidder is actually slow or just reading the wrong clock.