Ad Split Testing & AdWords Ad Rotation Settings

When it comes to managing AdWords accounts, split testing ads is a basic task on every account manager’s list. Split testing is a staple for consistent evolution and increased performance of any digital advertising medium really and is a must do for effective AdWords account management.

However, strategies vary and the ad rotation settings available in the AdWords tool are often confusing. The go-to ad rotation setting for most of the PPC community has long standing been the Rotate: Show Ads More Evenly option.

Rotate gives equal preference to all active ads in your ad group, regardless of how each ad performs. Each ad will be eligible to show an equal number of times, but the actual impressions — how often each is actually shown — can differ across ads based on a number of factors.

This option allows for multiple ad variations to run without intervention from automatic optimization from Google, this creates a controlled test environment and accurate data for decision making.

There are two other ad rotation options that provide automated assistance from Google in terms of optimizing toward different metrics.

Optimize for clicks (the default) gives preference to your ads that are expected to get the most clicks, based on your past clickthrough rates (CTRs). Google will try to show those ads more often than other ads in your ad group to help you gain more clicks and impressions.

Optimize for conversions gives preference to your ads that are expected to get the most conversions like purchases and sign-ups. Your ad group might receive fewer clicks, but it’ll likely receive more conversions. If there isn’t enough conversion data to determine which ad is likely to provide the most conversions, ads will rotate to optimize for clicks.

On split testing strategy includes the use of these options by cloning a version of the winning ad and running that ad (perceived as absent of any history) alongside a brand new version.

I do not recommend this strategy as it is not entirely clear that the cloned variation is really absent of previous history, and the AdWords definitions stat that they are working off of ‘expected’ metrics from more than just that ad variation. This is not a clean test.

To further complicate the matter Google has recently announced that the long standing Rotate: Show Ads More Evenly option will be changing.

Starting next week, the “rotate” setting for ad rotation will change. Instead of rotating creatives for an indefinite period of time, this setting will only rotate for a period of 30 days. After that, the setting will then optimize to show the ads expected to generate the most clicks. Every time a creative is enabled or edited, the ads in that ad group will rotate more evenly for a new period of 30 days.

This change will skew any split tests beyond the 30 day mark. This can be problematic for super granular or longtail campaigns as they do not always have the volume of data necessary to make an accurate decision. How much data is enough? Well, that is variable but you can use this PPC Ad Testing Tool to help you out. So, as you can imagine AdWords account managers everywhere are a bit peeved at this change, so much so that they have started a petition for change.

Hopefully Google will give us an option to opt out or a third party tool will come to the rescue with a solution. Until then we’ve just got to think outside of the adgroup. I’ll be setting calendar alerts to refresh ad tests every 2-4 weeks and will be looking into using AdWords Campaign Experiments for ads that need to marinate a little longer.