Ad rotation is the Google Ads setting that controls how impressions are shared when an ad group contains more than one active ad. It decides whether Google favours the ad it expects to perform best, or spreads impressions more evenly so you can compare ads yourself.
How ad rotation works
You set rotation at campaign level, and you can override it for individual ad groups. There are two options:
- Optimise: Prefer best performing ads. Google uses signals such as the search, device and location to pick the ad it expects to get the most clicks or conversions in each auction. This is the default.
- Do not optimise: Rotate ads indefinitely. Google aims to give each ad a more even share of impressions for an unlimited time, so one ad cannot dominate before you have judged it.
Two points limit how much the setting does. Rotation only decides between whole ads; within a single responsive search ad, Google always chooses the headline and description combination itself. And, at the time of writing (October 2026), campaigns using Smart Bidding are optimised by Google regardless of the rotation setting, because the bidding system already chooses which ad to serve.
Why it matters
Rotation affects whether your ad tests mean anything. With optimised rotation, Google may give one ad most of the impressions within days, based partly on early signals. That is efficient if you trust the system, but it makes a fair comparison hard: the losing ad may simply never have been shown to enough people.
Even rotation gives cleaner comparisons, but it costs you something. While the weaker ad keeps getting half the impressions, you are paying for clicks on a message you could have switched off. For a small UK account with modest traffic, that trade-off is real.
For most accounts today the bigger question is whether you need several ads per ad group at all. One strong responsive search ad, plus a second one testing a clearly different angle, is usually enough.
Common mistakes
- Setting “Rotate indefinitely” for a campaign on Smart Bidding and assuming it will be honoured.
- Judging an ad test after a few days and a few dozen clicks.
- Leaving rotation set to even after a test has a clear winner, so the loser keeps spending.
- Running several near-identical ads, which splits data without testing anything meaningful.
- Comparing ads on click-through rate alone when one brings better enquiries.
How to act on it
For most accounts, leave rotation on “Optimise” and run controlled tests through ad variations or campaign experiments, which split traffic deliberately and report on the difference. If you use manual bidding and want a straight comparison of two ads, switch that ad group to even rotation for the length of the test, then switch back.
Decide before you start what will count as a win and how long the test will run. A test needs enough conversions on each side to separate a real difference from chance, which for a small account can mean running for six weeks or more. Change one meaningful thing at a time, such as the offer or the main promise, rather than a word here and there, so the result tells you something you can reuse.
Judge the result on conversions and cost per conversion rather than click-through rate, and run whole weeks so weekday and weekend behaviour are both included. Planning and reading ad tests is part of my ongoing PPC management.