As the saying goes, “The only constant is change”.
You need to continuously test your campaigns to optimize performance.
And the easiest way to do that for Google Ads is to set up a campaign experiment. Experiments help you measure results so you understand the impact of your changes and make data-informed decisions.
Here’s a step-by-step guide on how to set up campaign experiments in Google AdWords:
Disclosure: Some of the links below are affiliate links. I will earn a small commission, at no extra cost to you, if you purchase through these links. Your support encourages me to continue blogging and help with the costs of hosting this site. Thank you!
1. Campaign Draft
Drafts are where you make changes to a campaign without impacting its performance.
When you create a draft, you’re mirroring your campaign. Make the changes that you want to test. If you’re unsure, you can leave and return to the draft when you’re ready to make changes. Or you can discard the draft altogether if you decide to run an experiment on another campaign instead.
When you’re done with the draft, you can create an experiment to test how this draft or variant performs against your original campaign.
1a. Choose An Existing Campaign For Your Draft To Mirror
Before you can create a draft, you need to have an existing campaign.
If you have multiple campaigns, decide on the campaign that you want to run an experiment on.
Also, think about what you want to test:
- Bidding Strategy
- Ad Copy
1b. Create Campaign Draft
To create campaign draft, click on Drafts & experiments on the left menu. Ensure that you’re on the Campaign Drafts tab then click on the plus button.
If you’re coming from All Campaigns, you’ll have to Select a campaign. If you’re coming from a specific campaign, you won’t see this step. The draft will automatically mirror the specific campaign.
Next, name your draft and add a description (optional).
The newly created draft will be listed under Campaign Drafts.
1c. Make Changes To Your Draft
Click on the new draft you created to open the draft and make changes. Any changes you make will automatically be saved. You can leave your draft and return to finish making your changes at any time.
I recommend that you don’t make too many changes. If you make too many changes, you won’t be able to tell which change led to the results. You can’t confidently attribute the results to a specific change.
Hence, do keep the changes to a minimum. For example, you can confidently say that target CPA resulted in more conversions at a lower cost per lead (compared to maximize conversions) since you only made one change—the bidding strategy.
2. Campaign Experiment
You can have multiple drafts for one campaign. However, only one of these drafts can run as an experiment at a time. You can schedule up to 5 experiments for a campaign so that they run one after another (since you can only run one experiment at a time).
When you’re done making the changes, you can convert your draft to an experiment. You will need to specify the start/end dates and the experiment split. Bear in mind that the experiment shares your original campaign’s traffic and budget.
Note that it will take some time for Google to approve your experiment’s ads. Thus, you may want to schedule your experiment to begin in the future.
Moreover, performance comparison only shows data for full days that fall between your experiment’s start and end dates. Rather than starting an experiment immediately, it’s better to schedule the experiment to launch tomorrow or in the future so you can compare all the data (if you start today, performance comparison will exclude today’s data since it’s not a full day).
When your experiment launches, either your original campaign or experiment is served.
2a. Choose An Existing Draft To Run As Experiment
You will need to have a draft to run as campaign experiment.
Just ensure that you’re done with the changes and that you’ve kept changes to a minimum.
2b. Create Campaign Experiment
To create campaign experiment, click on Drafts & experiments on the left menu. Ensure that you’re on the Campaign Experiments tab then click on the plus button.
Select the draft that you want to run as an experiment.
Name your experiment and add a description (optional).
Specify the start and end dates. In other words, how long you’d like the experiment to run. You can select None as end date if you want to manually end your experiment.
Next, decide on the experiment split. This is where you choose how much of your original campaign’s budget and traffic you’d like to use for the experiment. I usually choose 50% so the budget and traffic is divided equally between the original campaign and experiment.
There are Advanced options if you’re running an experiment for search campaigns.
- Search-based: Get statistically significant results faster. Users can see both experiment and original campaign if they search multiple times.
- Cookie-based: Users only see one version of your campaign, regardless of how many times they search. This helps ensure that other factors don’t impact your results and hence, giving you more accurate data.
Note that Google uses cookie-based split for Display campaigns so that users only see the experiment or original campaign.
Also, if you select the cookie-based split and use audience lists, ensure that you have at least 10,000 users in the list to run an experiment. Your results may be less accurate if you have less than 10,000 users.
3. Changes To Your Experiment
I recommend that you don’t make any changes after your experiment is launched. The results will not be accurate due to these changes that you make in the midst of testing.
However, I understand that there may be times where changes are necessary.
In such a situation, just click on the experiment under the Campaign Experiments tab and make your changes.
You can also choose to end the experiment by clicking on End Now at the top-right corner of the page.
If you click on the pencil icon beside your experiment name under the Campaign Experiments tab, you can edit the experiment name and description.
Double-check your original campaign and experiment after you make your changes. There are times when the changes are duplicated in the other campaign (which I didn’t want), probably due to the experiment mirroring the original.
4. Compare Performance
When your experiment is running, you can monitor and compare its performance against your original campaign. Or even choose to end your experiment early!
If your experiment performs better than your original campaign, you may consider applying your experiment to the original campaign. You can also convert your experiment into a new campaign with the same dates and budget as your original campaign and then pause your original campaign.
4a. Monitor Your Campaign Experiment
To view your experiment’s performance, just click on your experiment in the list of campaigns. There’ll be a beaker icon next to your experiment so you should be able to find it easily.
At the top, you’ll see a scorecard.
- Performance Comparison: Dates for which your experiment’s performance is being compared to the original campaign’s performance. Only full days will be shown.
Conversion-related metrics will only be available if you’ve set up conversion tracking.
Hover over each metric and you’ll see whether the data is statistically significant.
- Statistically Significant: Your experiment is likely to continue performing with similar results if it’s converted to a campaign.
- Not Statistically Significant: Many reasons for this including (a) not enough time to run the experiment, (b) not receiving enough traffic, (c) traffic split was too small hence experiment isn’t receiving enough traffic and (d) the changes you’ve made haven’t resulted in a statistically significant performance difference.
5. Conclude Experiment
Depending on the results, you can decide whether or not to apply the experiment.
If there’s no clear winner, you can try launching the same experiment with more budget or longer duration.
If your experiment performs poorly against your original campaign, it’s recommended to keep your original campaign running and end your experiment. You can try creating a new experiment to test something else.
If you’re happy with the results of your experiment, you can apply your experiment to the original campaign or convert your experiment into a new campaign.
5a. Apply Experiment
Look for the beaker icon then select the experiment from the list of campaigns. Click on Apply at the top-right corner of the page.
You will be given 2 choices:
Update Your Original Campaign
When you apply your experiment to the original campaign, your experiment will end and the changes in your experiment will be applied to the original campaign.
Convert To A New Campaign
When you convert your experiment into a new campaign, your experiment will appear in All Campaigns and run as a normal campaign. The original campaign will also be paused.
Now It’s Your Turn
Have you run any campaign experiments in Google AdWords before? What did you test? How was the performance?