There’s often a great debate over ad rotation. Should you use Google’s highly recommended Optimize or the ad tester’s preferred method Do Not Optimize?
We took a look at just a few ad groups for one of our advertisers to show them why ‘optimize’ is not working for them; and we’d like to share a few examples and pitfalls to watch out for when choosing your ad rotation setting.
#1 Google Makes Mistakes
The first thing to note in our charts, we show the ‘enabled date’ of the ad so you can see when an ad was first enabled. However, the stats are only based upon the timeframe comparison. So, in our first example while the one ad was enabled 5 days before the second ad; the data shown is the last 90 as both ads were active in that timeframe. We do not compare data for date ranges when some ads weren’t actively competing against each other.
In our first example, the ad with the best CTR (which is not statistically significant), conv. rate, conversion per impression; and every single other metric was only served 29,866 times versus the ad that has the worse metrics was served 20 times more often with 617,723 impressions.
This is an obvious example of Google just making a bad choice.
#2 Google Treats New Ads In Strange Ways
When you create a new ad and you are using optimize, you usually find one of two things true:
- The new ad never gets a chance to compete
- The new ad gets preferential treatment
Yes, these are two widely varies options. It sometimes seems random when you create a new ad if the new ad will just take over all the impressions or just never get a chance.
In this ad test, a new ad was created on August 14 in an ad group where the established ad had been running almost a year. The new ad has significantly lower CTRs, conversion rates, cost per conversion, and is pretty much a failure stat wise.
However, Google has shown the failed ad more often than the much better ad (42,428 impressions vs 27,875).
In this case, the opposite is true; Google has barely displayed the new ad (which is in the same campaign as our previous example). While it does have a lower CTR, it has a higher conversion rate and conversion per impression. With only 4 clicks; it’s impossible to tell if those trends will continue or not but when your first 4 clicks turn into 5 conversions; you probably want this ad to have a chance.
Yet this new ad has only been displayed 82 times in the last 7 days while just one ad has dominated the impressions at more than 2,000. In addition, when our new ad shows, its average position is 2.9 compared to our highest impression ad at a 1.6; therefore, Google isn’t even treating these ads evenly to figure out the best one to show.
It is worth noting that our middle ad does have a better conversion rate and conversion per impression than our top impression ad. Optimize isn’t really showing the top ad in terms of how many conversions we can get for this ad group – it’s chosen this highest CTR and not the best ad for our business.
#3 When the Metrics are Mixed; Google Falls back on CTR
In this ad test, there’s one ad that’s a winner by CTR and another ad that is a winner in every other metric. If you were truly being optimized, then the ad with the highest Conversion per Impression should be the ad shown the most often.
In this case, the ad with the highest CTR is being displayed 3 times as often as the ad with the highest conversion rate and conversion per impression. When you have one ad with a higher CTR than another and yet another ad with a higher conversion rate or conversion per impression; Google often defaults to CTR over other metrics.
#4 Sometimes Google Picks Winners for No Apparent Reason
In this ad test, we have a clear winner with a higher CTR, conversion rate, and conversion per impression than the other ads in the ad group. However, Google has decided to show that ad less often than second best ad in all of these metrics.
#5 When You Create 5+ Ads Per Ad Group – Google Just Seems Confused
Google often recommends to just keep creating ads. There’s a lot of advice these days on just keep making ads and never pause any.
When that happens; we often see stats like this:
- The ad with the highest impressions is rarely a winner in CTR or conversion per impressions
- The ads with the lowest impressions are usually the lowest CTR ads
- The ad with the highest potential to convert (either conversion rate or conversion per impression) is shown often; but not as much as the random ‘winner’ of most impressions.
In this ad group; the first ad is the clear overall winner in terms of conversion per impression and it’s the second highest CTR. It’s served much less often than the ad with the highest cost per conversion (which has the most impressions) and the ad with the 3rd highest CTR; Google isn’t even falling back to CTR.
There are many examples of loading up on many ads causes ad serving confusion.
#6 When the Data is Obvious – Google Usually Gets it Right
Usually, but not always, when one ad is a winner in every single metric – Google will show that ad the most often (although, one could argue that ad 2 is still shown too often with such a clear trend):
As soon as metrics are mixed (one ad is a winner in CTR and another in conversion rate); Google often shows the higher CTR ad more than the other ads even if the other ads have higher conversion rates or conversion per impressions.
When you introduce 4-5 or more ads; the ad serving seems more random than based upon actual metrics for ads that are not clear losers in the ad group.
New ads either get ignored or preferential treatment. If you are adding a lot of new ads – watch to make sure the new ones get a chance to compete.
Overall rotation thoughts:
- If you are serious about ad testing, use rotate
- If you are adding many new ads, use rotate until the ads get enough data to where you can turn on optimize and let Google figure it out for you
- If you are ignoring your ad tests and not adding many new ads, then optimize can be OK to use as Google making some good and bad choices is better than doing nothing
If you’d like to automate these stats and easily choose winner and loser ads; take a look at Adalysis.com.