How to Test AMP Pages with Google’s Cookie-Based Splits

Last updated on by Brandon Weaver in AMP, A/B Split Testing, Google Ads

Google recently released a new feature to test AMP pages, allowing you to conduct cookie-based A/B testing:

As many digital advertisers know, AMP pages can have a big impact on mobile conversions because with faster loading pages, fewer people bounce due to the mobile experience. Yet, you may still want to test AMP pages against traditional web pages or even against other AMP post-click landing page designs.

You can do both with Google’s search-based or cookie-based testing. In this article, you’ll learn what cookie-based splits are, how they differ from search-based splits, and how to set up cookie-based testing in Google Ads.

What are cookie-based splits?

Previously, to A/B test AMP pages your only options were to use campaign experiments (that utilized search-based splits) or ad variations.

Using campaign experiments with search-based splits would show a Google user a different page every time they conduct a new search. For example, on their first search, a person may see your “control” page and on their second, they may see your “test” page. These experiments operate the same on the Display Network. If a person lands on a page for the first time, they may see your control page whereas as if they refresh or revisit the page, they may see your test page.

With Google’s new cookie-based splits, Google users will only see either the test or control page. So, if on their first search a person sees your test page, they will continue to see that page over and over no matter how many searches they conduct (during your experiment).

Benefits and disadvantages of each AMP testing method

You still have two main types of A/B testing inside Google Ads:

Search-based splits and cookie-based splits are different testing options available to use within campaign experiments. Each method has its own benefits and disadvantages detailed below.

Campaign experiments

Campaign experiments work alongside campaign drafts in that they allow you to copy your current campaign’s setup and from there, make any desired changes. Once you’ve finished your draft, instead of applying the changes to your original campaign, you can set the draft up as an experiment.

When using campaign experiments, a Google user may see your original campaign or your experiment, depending on how you’ve decided to the split the traffic share.

If you make any changes to your experiment campaign, those changes will not be added to your original campaign (and vice versa).



As mentioned earlier, cookie-based splits and search-based splits are two testing options within campaign experiments. Using cookie-based splits can help ensure that other factors don’t impact your results.

Which method you choose is up to your discretion. However, cookie-based splits may be a better option if you’re testing a specific audience or want to create a post-click landing page test that runs similar to other CRO testing tools.

Search-based splits may be a be a better option if you want to maintain consistency with previous search-based tests you’ve already run or you are testing low-volume campaigns.

Using ad variations

Ad variations are useful if you want to test variations of specific ads across multiple ad campaigns or your entire account. As an example, you may create a search ad with the call-to-action text “Buy now” and another with the call-to-action text “Buy today.” Then, you can test which ad variation performs better.



So which testing method should you use?

How to set up campaign experiments with cookie-based splits

If you plan to test an AMP page against a standard web page, make sure that both pages look as similar as possible. Ideally, the design, copy, and layout would be exact on each page. Make sure to consider differences in text size, content, formatting, images, and submission fields. Differences in any of these areas could skew your test results.

With both pages identical, follow the steps below to set up your campaign experiment with cookie-based splits:

Step 1
Log in to your Google Ads account and click “Drafts & experiments” from the left-hand menu. From this page, click “New Draft” to create a campaign draft:

Step 2
Choose the campaign to use for your experiment and name your draft:

Tip: Clearly identify your campaigns to make them easier to find later. For example, “Campaign name_AMP.”

Once you’ve selected a campaign and named your draft, Google will duplicate the original campaign. From there, you will need to go through each ad and keyword that has a final URL defined and update it with the appropriate AMP URL.

Step 3
With that complete, you can create an experiment. In the left-hand menu, click “Draft & experiments” and then “New Experiment:”

Step 4
Next, name your experiment, set start and end dates, and set the experiment split:

The experiment split is the percentage of your campaign’s budget to allocate with the experiment. Google recommends a 50/50 for the quickest test results.

(Note: You cannot change the budget split once the experiment has started.)

Step 5
Click “Advanced options” to reveal the experiment split options, search-based splits, and cookie-based splits:

Select “Cookie-based” and click “Save” to begin your experiment.

Build AMP post-click landing pages quickly and efficiently

Google’s new cookie-based testing offers yet another way to test AMP pages against standard web pages and realize the performance benefits of using AMP.

With cookie-based testing, you can make sure Google users only see either your test or control page throughout your experiment. This method also helps eliminate other factors that could impact your test results.

However, regardless of the testing method, you still need to create AMP post-click landing pages to provide the best mobile experience.

This is where Instapage’s AMP builder can help you create pixel-perfect post-click landing pages in minutes. To see the AMP builder in action, request a demo today and experience the Instapage difference for yourself.

Turn More Ad Clicks into Conversions

Try the world's first Post-Click Automation™ solution today. Start a trial or schedule a demo to learn more about the Enterprise plan.