AdWords Experiments is a tool to use when you want to split-test campaign changes without harming your original campaign. Simply stated, it’s a game-changer.

How it works is you first create what AdWords calls a "Draft". A draft is basically a clone of an existing campaign with all of its settings. The "Drafts" button is on the top-right of the screen.

adwords1

Click "Create new" and give your draft a name. 

adwords2

On the left navigation in AdWords, find "All Drafts".

adwords3

This is where you can see your draft (clone) campaign. In this draft campaign, you can now make experimental changes to your original campaign. None of the changes you make here affect anything yet so make the changes to reflect the hypothesis you are trying to test.

For example, let's say we currently use a "Manual Bidding" bid strategy where we manually make keyword bid changes. If our goal is to maximize the number of clicks, one experimental change we might consider is to use AdWords "Maximize Clicks" bid strategy to see if we can get better overall performance and more clicks than our current manual bidding strategy.

In your draft, you would just navigate to the Settings tab, find the "Bid strategy" section and choose Maximize clicks. 

adwords4

Make any other changes you want to test. When you're finished, look for the blue "Apply" button in the top-right. Select "Run an experiment" and click "Apply". 

adwords5

Here, you give the experiment a name (ex: Maximize Clicks Experiment – Campaign ABC), define how long the experiment will run, and decide how much traffic to allocate to it.

adwords6

Give it some time and you will soon see your experiment under the "All experiments" sidebar section in AdWords.

adwords7

As the experiment is running, you will be able to check in on how things are going and compare the experiment results to your original campaign performance. AdWords also calculates the statistical significance of the changes (the tiny little up and down carrots next to the numbers).

adwords8

When the experiment is completed, you can update your original campaign with the experimental changes, or pause your original campaign in-favor of continuing onward with the the experimental campaign.

adwords9

A few major learnings to take away from running experiments:

  • When you launch an experiment, the experiment is a new campaign! This will be important to know for budget management, reporting, and consolidating.
  • The metrics you see in the "All campaigns" section in AdWords won't include metrics from your experiment. That's because experiments are actually completely separate campaigns (mentioned in point 1 above) but they do both share the same Base Campaign budget.
adwords10
  • Experiments can't be used with shared budgets. This can be a barrier for some people that have to manage many campaigns and budgets in one account.
  • If all you're testing is a really simple change, it might make sense to just change your original campaign with those changes instead of applying the experiment to the original campaign. In our example, we only changed the bid strategy from Manual to Maximize Clicks. Therefore, if we find that Maximize clicks works better, we can simply make this change in our original campaign and not worry about applying the change through the Experiments section. The reason is, you might have made some changes to your original campaign since you made your draft (ex: adding sitelinks). You will lose those changes if you did not also make those changes in the Experiment.

If you have any questions, let us know. We're happy to help! 

Categories: