Branded search campaigns: paying for clicks you already get for free?

Branded search looks great in the report and means less than it seems. If you already rank first organically for your brand name, a paid click often just replaces a free one, so the spend is closer to a tax than to acquisition. On the other side, competitors bid on your name, the results page is crowded with ads and AI overviews, and if you don't run an ad, a rival can end up sitting above your organic listing.

My view is to keep branded running but cap the budget and check it properly. A pause test is the best way to do that. Say a plumber switches branded off for two weeks and total calls barely move. That tells you more than any last-click number ever will. How would you run that test without losing too much in the meantime?

Honestly, it comes down to whether anyone else is bidding on your name. If no competitor touches your brand terms and you hold the top organic spot, pausing mostly loses clicks you'd have gotten for free, and call numbers may barely move. That's the result Atlas is describing. But if a rival is bidding, pausing hands them those searches, and your test ends up measuring their gain as much as your loss. You can check this before you pause: the auction insights report for your brand terms shows who else is in the auction.

For the test itself, I'd avoid switching branded off across the whole account for two weeks. A seasonal dip or a promo in that window will look like a branded effect. Splitting by region or by time block gives you a cleaner comparison, and you can compare calls per branded impression between the groups.

I don't think calls per branded impression works here. Once you pause a region, its branded impressions go to zero, so the ratio is meaningless on exactly the side you're trying to measure. Compare total calls per region against a baseline instead. And keep an eye on branded organic clicks in Search Console, because that's where any substitution would show up.

Agreed, that ratio breaks the moment the denominator hits zero. I'd track total branded clicks per region instead, paid and organic combined. If the paused regions pick up organic clicks that roughly match the paid ones you lost, that's substitution and the spend was mostly a tax. If the total drops, the ads were pulling demand you wouldn't have gotten otherwise. Search Console alone can miss part of this, since some of the gain may show up as direct traffic or as people searching the name again later.

You also need a control for the region split. Regions differ in size and seasonality, so a plain before and after comparison is shaky. Keep one region running as a control and compare each paused region's change against the control's change over the same weeks. Without that, a local news story or a holiday can make the result look real when it isn't.

Are your call numbers tracked by location, or only as account totals?

Set the pause length to your usual lag from first branded click to booked call, plus about a week. Branded searchers often come back days later, either by typing the URL straight in or by searching again. So a two-week test on a service with a ten-day consideration cycle will miss part of that recovery, and the spend ends up looking more valuable than it really was. Pull the lag from your call and conversion data before you pick the dates.

Elena's lag point is the right one, and there's a filter I'd put in front of it. If you rank first organically for your name and no competitor bids on it, most paused clicks just move to organic, so a long test mostly tells you how good your organic listing is. In that case I'd cap the branded budget and skip the test. Pause only when a rival is bidding or your organic position is shaky, because that's the only situation where the result would change what you do next.