Location targeting for a local clinic: presence only, or the default too?

Picture a dental clinic in a mid-sized city. Google's default location setting covers people who are in or regularly in the area, plus anyone showing interest in it, and for a clinic that sounds about right. The catch is that plenty of clicks from the wider metro come from people searching at work or visiting family, and they rarely book. Switching to presence only cuts reach hard, and you lose the patient who lives twenty minutes out and would happily drive in. Distance matters less than people think for dental work, so I'd keep the default. Then trim the waste by excluding the far suburbs or the areas that never convert, instead of shrinking the whole region. Has anyone seen the cheaper out-of-area clicks convert at a noticeably lower rate, or do the numbers come out about even?

Agree on keeping the default, but I'd settle it with the location report, not the campaign totals. Break conversions out by matched location and location type (presence vs. area of interest), then compare cost per booking by city or postcode cluster.

Make sure the conversion you're judging is a booked appointment or a tracked call, not a form fill. People from far out fill in forms more casually and then never show up, so form fills will flatter the outer areas.

If the outer suburbs cost roughly double per booking, exclude them. If they're close to the core, the default is earning its keep.

My rule is to leave an area alone until it has around 10 booked appointments. Outer suburbs usually send a trickle of clicks, so a swing of two bookings either way can move cost per booking by 50% or more. Below that number, look at the trend over a longer window instead of a snapshot. Once an area clears 10 and still runs at roughly double the core, that's when excluding it makes sense.

The 10-booking threshold makes sense, Elena. The setting itself is a bit off, though. Google Ads defaults to "People in or regularly in your targeted locations," which is presence only. The broader option, "in, or who show interest in," is what pulls in the work-search and visiting-family clicks Marco described. So if the clinic is still on the default, it's already on the narrow setting, and the real test is whether moving to interest adds bookings at a sane cost. I'd also split the location report by location type, since the cheap far-out clicks usually land in the "area of interest" rows.

Presence only is the default, so the interest option is the one to test. Flipping it live won't give you a clean read, though. Volume jumps and bidding has to relearn, so the before and after aren't comparable. A campaign experiment or a duplicate campaign with a split budget lets you compare the same weeks side by side. Otherwise the 10-booking rule will take a small clinic most of a year to reach.

Agreed on running an experiment, but a duplicate campaign with the interest setting will overlap with the original in the core city. Both end up bidding on the same searches from the same people, so unless you exclude the core from one of them, the split isn't clean. A proper campaign experiment divides the traffic for you, which makes it the better fit here. And allow for lag: a dental booking can take a week or more after the click, so hold off reading results until the conversion window has closed.

Agreed on the experiment. Because the split is by user, the interest arm's extra clicks come from new people rather than the same ones getting reshuffled, which is the comparison you want. Judge the arms on cost per booking, not total bookings. The interest arm will almost always win on volume and lose on efficiency. For a small clinic, a 50/50 split can take months to reach ten bookings per arm, so giving the control the larger share is a reasonable compromise.

The 80/20 split makes sense. The thing to watch is bidding. With Target CPA or Maximize Conversions, each arm learns on its own, and the interest arm will often look cheap for the first couple of weeks because it's still spending its way into a bid that suits its new clicks. A short early read can flatter it. So run the experiment for a fixed window and only compare cost per booking once both arms have finished learning. With manual CPC or a fixed bid this matters much less, but with automated bidding the first few weeks are the least representative ones.

Agreed, and the easiest mistake is giving the two arms different target CPAs. If the interest arm gets a looser tCPA so it can chase volume, you end up testing the bid target instead of the location setting. Set both arms to the same number before the split starts.