For the last few months, Google has had an exclusive beta kept underwraps for select accounts at Seer. The beta is a new bidding strategy called Maximize Conversions.
The beta was initially only available to run as a 4-week experiment within a select number of campaigns that Google chose based on a specific set of criteria.
When Google first introduced this exclusive beta to only four accounts at Seer, we asked what factors they were using to determine which campaigns could test the beta. There were two main pieces of criteria that we could get our hands on:
- If your campaign generates at least 15 conversions/week
- If your campaign is consistently hitting its daily budgets
One thing to note is that the control in this experiment was whatever other bidding strategy your campaign is currently running. For example, if the campaign that you are running the experiment within is currently using the bidding strategy Optimize For Conversions, then that will be the control in the experiment.
Since being offered this beta, Seer has been able to run over 10 experiments within various industries.
As of earlier this week, the bidding strategy is now out of beta and available to all accounts/campaigns at Seer and will be rolled out fully to all advertisers by end of Q2.
Below I will take you through the lessons we learned from the experiments we ran & recommendations for when to test this new bidding strategy in your accounts.
Lessons Learned & Recommendations
Lesson #1: It takes about 2 weeks for Google to learn how to optimize the campaign.
- For several of the campaigns that we ran this test in, we saw performance take go down slightly those first two weeks then head up into the right for the remaining two weeks.
- Recommendation: Don’t pause the experiment two weeks in once you see that performance is down. Let the experiment run for the whole 4 weeks.
Below is an example of one B2C client that saw performance down after the first two weeks but then pick back up the last two weeks.
First two weeks:
Last two weeks:
- At Seer, we ran 10+ tests and of the test, about 15% of the time, the experiment performed worse than the control, which was the campaign original bidding strategy.
- Recommendation: ALWAYS run this as an experiment first at a 50/50 split before rolling out 100% to your campaign. There is no consistent theme in terms of what types of campaigns this experiment did not perform well in, therefore, you should always run an experiment first.
Below is an example from one of our B2C clients where the experiment performed worse than the control
No one industry performed better than another. Seer was able to test this for education, a medical device company, healthcare, and a decking material company. Each of these industries saw at least one very successful test with this bidding strategy.
- Recommendation: Run an experiment! We have yet to find an industry that this experiment does not work for at all.
Medical Sales Client
- Since the control in the experiment is the original bidding strategy (rotate evenly, rotate indefinitely, etc), Seer recommends that the original bidding strategy have already run for 30 days. If not, then you could run into the issue that Google is trying to optimize for two different bidding strategies at the same time rather than focusing on the new maximize conversions strategy.
- If there is one thing that you should take away from this entire post, it should be that you should always run this new bidding strategy as an experiment with a 50/50 split before rolling it out 100% to your campaign. This will allow for a completely even test and determine a clear winner at the end of the 4 week experiment.
Enjoy the conversions!