Google Ads offers its users the ability to create drafts and run experiments on their existing campaigns. There are some limitations to this, as you must be running either a Search or a Display Network campaign. With drafts, users can propose and draft up multiple changes to a single campaign, without actually altering the original. Experiments are ideal if you’re wanting to revise your current strategy and run it by your team or a client for approval before taking action. Sure the new strategy you drafted up looks good on paper, but will it outperform your original campaign?

Thanks to Google Ads’ experiment feature, we can test and see just how well our drafts will perform. Once you have completed drafting your modified campaign, you’re given the option to either apply the changes directly to the original campaign or convert your draft into an experiment. When running an experiment campaign, you can choose how long you run the experiment for (It’s recommended to run them for at least 4 to 6 weeks to reach statistical significance) and how much of your original campaign’s traffic and budget you want to use. 

When your experiment campaign is live, it will run simultaneously alongside your original campaign. When a potential client uses google search or visits a webpage on the display network, either your original campaign or experimental campaign will randomly be made available for the auction. Although still random, the choice between showing the original campaign and the experimental one is highly influenced based on how you’ve split the traffic share between the two. If your campaign experiment is underperforming, you can choose to end it early. If your experiment has run for less than four weeks or if your traffic split was too small, your experiment might never reach statistical significance, without this, it’s hard to tell just how well the experiment will perform if converted into a campaign.

Once you have completed your experiment and are happy with the results, you can apply it directly to the original campaign, or you can convert it to a new campaign, pausing the original in the process. The newly created campaign will appear with your other existing campaigns and have the same budget and dates as the original. Whatever route you choose to take after completing your experiment, it’s essential to know that performance data for both the original campaign and the experiment will be preserved.