To compare two responsive search ad copy directions on clicks, both ads have to get a fair share of impressions. Google's default ad rotation, Optimize, serves the ad it expects to get more clicks more often, so it can pick a winner before you've read the test. Switching the ad group to Do not optimize makes the ads enter the auction more evenly.
There's a catch. If the campaign uses Smart Bidding, Google Ads automatically uses Optimize, whatever you select. Target CPA and Maximize conversions both count. On those campaigns, the clean way to test copy is an ad variation in Experiments.
Where the advice came from
On a mentoring call, a marketer running Search ads for a software product wanted to know which message their audience responded to: an ad that names the product category, or one that sells the outcome. Our advice was to keep the keywords the same, change one or two headlines, turn off optimized rotation and let it run for one to two weeks. Then compare clicks and CTR, since both ads would be answering the same searches.
That advice holds on a manual or non-Smart bid strategy. Their campaign was on Target CPA, though, and earlier on the same call we'd suggested moving it to Maximize conversions. Both are Smart Bidding, so the rotation setting wouldn't have evened anything out.
What the two settings actually do
- Optimize. Google optimizes for clicks in each auction, using signals like keywords, search terms, device and location. As data builds up, the ads it predicts will do better get served more often (and on Smart Bidding, it predicts conversions too). Serving can stay fairly even when the ads perform alike or the ad group gets little traffic.
- Do not optimize. Ads enter the auction more evenly, for an indefinite amount of time. Google doesn't recommend it for most advertisers, and impressions still won't be perfectly even: a higher quality ad can show on page one while a weaker one lands on page two.
Ad rotation is available in Search, Shopping and Display campaigns. Set it at the campaign or ad group level under Additional settings, then Ad rotation.
Rotation works between ads in the same ad group. On the call, the two messages sat in two ad groups with the same keywords. That doesn't split evenly either: when keywords in two ad groups can match the same search, only one triggers an ad, chosen by Google's ad group prioritization. Put both versions in one ad group. Google recommends at least two responsive search ads per ad group anyway.
On Smart Bidding, test with ad variations
Ad variations live under Experiments, and they're available for Search campaigns. Choose responsive search ads as the ad type, then either find and replace a phrase or update the text: add, remove or pin headlines and descriptions. You set the experiment split, and Google splits users by cookie, so each person sees only one version however many times they search.
The report compares the variation with the original. Each metric shows the difference with a confidence interval (80% by default, and you can change it), and a blue asterisk marks a result Google rates as statistically significant. When it's done, you can apply the winning text to your ads.
Whichever method you use, change one thing. If both headlines and descriptions change at once, you won't know which one moved CTR. We also look at conversions before rolling a winner out, because the ad that earns the click isn't always the one that earns the lead.
The takeaways
- Optimize rotation favors the ad Google expects to get more clicks, which can settle a copy test early.
- Do not optimize enters ads into the auction more evenly, but quality can still make impressions uneven.
- On Smart Bidding, including Target CPA and Maximize conversions, Google uses Optimize automatically.
- Keep both ads in one ad group. Rotation doesn't balance ads across ad groups.
- On Smart Bidding, run an ad variation in Experiments and read the confidence interval.
Common questions
What does Do not optimize ad rotation do in Google Ads?
It makes the ads in an ad group enter the auction more evenly, for an indefinite amount of time, instead of favoring the ad Google expects to get more clicks. Google says it isn't recommended for most advertisers, and impressions may still be uneven because a higher quality ad can show on the first page while a lower quality one shows on the second.
Does ad rotation work with Smart Bidding?
No. Google says that if you use Smart Bidding, Google Ads automatically uses the Optimize ad rotation setting. Smart Bidding includes Maximize conversions, Target CPA, Maximize conversion value and Target ROAS, so on those strategies choosing Do not optimize won't even out delivery.
How do I A/B test responsive search ads on Target CPA or Maximize conversions?
Use an ad variation. In Experiments, choose Ad variations, target responsive search ads, then find and replace text or update headlines and descriptions. You set an experiment split, and Google splits users by cookie so each person sees one version. Results show the difference with a confidence interval, and a blue asterisk marks a statistically significant result.
Where do I change the ad rotation setting?
At the campaign or ad group level. In Google Ads, go to Ad groups, open the settings for the ad group or campaign, choose Additional settings, then Ad rotation, and pick Optimize or Do not optimize. You can also change it for several campaigns or ad groups at once with Edit, then Change ad rotation.
Can I test two ads in separate ad groups with the same keywords?
Not cleanly. Ad rotation controls how ads in the same ad group are served relative to one another. When keywords in two ad groups can match the same search, only one of them triggers an ad, chosen by Google's ad group prioritization, so put both versions in one ad group.