Emergency plumber ads: run them overnight or only when someone picks up?

Take a plumber who wants ads running around the clock because emergency searches happen at 2am. The catch is what happens after the click. If nobody answers until 8am, you're paying for a lead that calls the competitor who picked up instead.

My view is to keep emergency keywords on overnight only if there's a real person or reliable on-call answering behind them. Otherwise, limit the schedule to staffed hours.

I wouldn't switch off-hours off completely on day one, though. Put after-hours traffic in its own ad group with a negative bid adjustment, so you can see how many of those calls actually turn into jobs once the team is back. Has anyone measured that gap with call tracking, or do most people just guess?

Marco, an ad group can't hold a time-of-day rule. Ad schedule bid adjustments live at the campaign level, so moving the after-hours keywords into their own ad group won't split the traffic. To measure the gap, put the emergency keywords in a separate campaign with its own schedule and a negative adjustment, then compare its call-to-job rate against the daytime campaign. Call logs are timestamped, so you could also filter a single campaign by hour, but a separate campaign is easier to read later.

Agreed, the call-to-job gap is the number that matters, so filtering by timestamp is a good shortcut. One thing to watch when you split by hour: tag each call with the time it started, not the time it was answered. Otherwise your 2am calls will look like they got picked up at 8am.

The click time matters too, since ad schedules key off when the click happens and jobs often get booked hours later. If you attribute a 2am click's job to the booking date, the after-hours gap looks worse than it really is. Which timestamp does your call tracking use by default?

Iris is right, and I'd add that after-hours calls rarely turn into jobs on the call itself. Most go to voicemail and the callback happens at 8am, so the booking lands in a different call log, often without the original click attached. If the callback and the booked job don't carry the gclid from the first click, the job gets credited to daytime traffic and the after-hours gap disappears. Keep the gclid on the lead record through the callback and the 2am cohort will show its real rate.

Agreed on the gclid, but whether it survives the callback depends on where the overnight voicemail goes. If it lands in the same CRM record as daytime leads, carrying the click ID through is easy. If it sits in a separate inbox or call tool, the 8am callback usually gets created as a brand new lead with no click data, and you can't rebuild that link afterward. So before trusting the 2am conversion rate, make sure the first contact and the callback share one record. If they don't, capture the gclid at the moment of first contact, because nothing downstream can recover it.

Nova's point about the shared record decides whether this measurement is worth doing at all. My rule of thumb: if the overnight voicemail and the 8am callback don't live in the same record, treat the after-hours campaign as unmeasurable and don't cut its schedule based on conversion data. Until the records are joined, judge it on something else, like how many of those calls reach a human the next morning. Otherwise you'll read a missing link as a bad keyword and switch off traffic that was actually working.

Same trap on Microsoft Ads, where the click ID is msclkid and offline conversion import only matches the callback if that ID is stored on the record. Also check the minimum call duration on your call conversion, since whether a voicemail call counts depends entirely on that threshold.