The skills we teach deliver real ROI.

Using Experiments in Google AdWords

Using AdWords Campaign Experiments, you can test changes to your AdWords account based on your “What if” questions. This allows you to get a sneak preview of the outcome before you fully implement changes to your account.

With experiments, you can test changes to keywords, bids, landing pages, and ads. What you cannot test are campaign level changes, such as budgets, location targeting, or scheduling. Although you can test dynamic keyword insertion, it is not a great for Experiments. Long keywords will not fit and short ones may waste space in the ad.

Some people test by making a change to the whole account, such as increasing keyword bids, and seeing how the campaign performs. Once that change is made, the CPC, CTR or other metric could go up or down. The problem is that with this approach, you cannot know for sure that changing the keyword bid caused that change. There could be other factors at play, such as similar advertisers entering or leaving the auction. It could also be that the content of your ad is trending which could increase clicks.

With an Experiment, the account owner can test this keyword bid change first. When that is the only thing changed in an AdWords account, the difference in performance is compared to not changing it in the control group. This way, account owners can see if the change in metrics was a result of the higher keyword bid. Otherwise, there is no way to know if the change occurred as a result of the bid adjustment or if there was another unknown factor.

In addition to keyword bids, testing ad copy is another good use for Experiments. One test could be to have ads that show price compared to ads that do not show price. You can see the difference in the performance very clearly before these two groups. For those sites that are not eCommerce, they can test ads by making changes to the header only. There are a number of factors that can be changed in an ad.

Be careful when you get started with this tool. There are a lot of options for testing and you can test more than one thing at a time in an Experiment. However, you are much better off testing only one thing. At the end of the experiment, it will be much easier to determine why something performed better. On the other hand, if you do decide to test a number of different factors in one Experiment, it will be much more difficult to determine why the results were positive or negative.

Experiments have the same costs as regular campaigns. If what you’re testing is a raise in bid, you’ll want to keep a close eye on your account and consider changing your daily budget during the length of the Experiment. Experiments also have an impact on the quality score. If it turns out the change made during the Experiment negatively impacted the quality score, there is really no need to worry since this is a short-term test. Just be aware of it.

What if you’re not sure where to even start? Look at the suggestions from the Opportunities tab in your AdWords account. A mistake some people make when managing an account is blindly accepting all suggestions from this tab. These suggestions should be reviewed before they are applied to an account because some are not relevant to your marketing goals. However, this tab can still be useful to give you some ideas on what to test.

Google’s recommendation is to run the campaign for two to four weeks. While the campaign is running, you can view the results by segmenting on Experiment. After the Experiment has run for a few weeks – or long enough to collect meaningful data – you can stop the experiment with no permanent changes made to the account. If you’re not sure you have enough for it to be meaningful, you can use to quickly see the significance of the CTR differences between your ads.

If you find that conversions increased during the Experiment but there was not a large increase in the cost for each conversion, you may want to use this bid all the time. If the results are positive however you measure it, you can apply the changes that were tested in the Experiment knowing that you did your homework first and are not just blinding applying bids. Most importantly – keep track of what you’re doing! Overtime, you may have tested dozens, or hundreds, of things in your account. If you don’t track what worked and what didn’t, you may be conducting the same test in a short period of time.

Keep these factors in mind and Happy Experimenting!

Leave a reply