Test your AMP campaigns

Faster landing pages typically lead to more conversions, and Accelerated Mobile Pages (AMP) let you create landing pages that load very quickly (many load in less than one second). Combining speed and smoother loading, AMP landing pages often provide a better landing page experiences, and landing page experience is an important factor in your Quality Score.

This article will show you how to test if your AMP landing pages are driving more conversions. An A/B test is a way to compare two versions of a single variable (in this case, your landing page) to see which version is more effective.

Ways to A/B test your AMP campaigns

There are two different ways to A/B test your AMP campaigns. The testing method you choose will depend on the level of detail and structure of your campaigns.

Method Benefits Things to consider
Drafts and experiments
  • Allows you to test multiple changes on a single campaign
  • Provides Google Ads Editor support for users to create and edit experiments in bulk changes
  • Since you can only launch experiments on a campaign-by-campaign basis, you should use this method with campaigns that get a lot of mobile traffic.
  • After you launch the experiment, you’ll need to ensure that any changes made to the original campaign are also made to the experiment campaign (or vice versa).
Ad variations
  • Allows you to test AMP pages across multiple campaigns.
  • Launch across multiple campaigns with one click.
  • Updates original and modified campaigns when you make a change to the original.
  • No need to create duplicate campaigns for testing.
  • There needs to be an entry in the Mobile URL field that specifies a variation on the URL string. Your URLs also need to follow a consistent pattern for string replacement (for example, m.example.com / amp.example.com).
  • Does not allow for bidding changes; only allows for creative and URL changes.

Before you begin

Here are some tips to help your AMP test campaigns run smoothly:

Do: Use https://search.google.com/test/ampto make sure your AMP page is valid during the test period.
Don’t: Start a Drafts and Experiments test immediately upon launching your AMP landing page. Please wait 2 days to ensure your AMP page is served from cache before testing.

Do: Run experiments on campaigns that target mobile traffic.
Don’t: Run experiments on campaigns thathave sitelinks, as they aren't currently supported by AMP cache.

Do: Use cookie-based splits to make sure users see only AMP or non-AMP pages no matter how many times they search. Since AMP pages are served from cache only in this format, use only for text ad-based Search campaigns .
Don’t: Place parameters that are used purely for tracking purposes before the {ignore} parameter in the mobile final URL. They should come after the {ignore} parameter.

Do: Schedule and run the test for at least 2 weeks, or until D&E shows statistically significant results.
Don’t: Use campaigns that have less than 250 clicks per day.

Do: Focus your tests on one variable at a time. If you want to examine the effects of more than one change, use separate tests.
Don’t: Avoid making changes to campaigns while running your test. If you need to make changes to your AMP page, mirror those changes in your test campaigns too, as soon as possible.

Do: Use for search text ad campaigns because AMP pages are served from cache only in this format.
Don’t: Forget that creatives in experiment campaigns go through ad approvals, just like those in other campaigns. This can take about a day, so keep this in mind when setting your experiment start date. Similarly, if you choose to add experiment creatives back to the original campaign, those creatives will also need to be reviewed.

If you plan to run any type of experiment with a campaign that uses Google automated bidding, allow at least one week for both the original campaign and experiment campaign to recalibrate to their new traffic levels before assessing results.

To make sure your test results are as accurate as possible, check that your AMP and non-AMP landing pages look exactly the same (or as similar as possible). Watch out for differences in:

  • Text (size, content, and formatting)
  • Images (size, content)
  • Submission fields (number of fields, pre-populated text, date pickers)

If any of these elements are different between the AMP and non-AMP page, then they may skew test results and paint an inaccurate picture of AMP performance.

You can't create experiments on campaigns that have shared budgets or test AMP with Dynamic Search Ads (DSAs).

Running an experiment when using Google auto-bidding

If you plan to run any type of experiment with a campaign that uses Google automated bidding, allow at least one week for both the original campaign and experiment campaign to recalibrate to their new traffic levels before assessing results.

Running tests with Search Ads 360

Drafts and experiments cannot be created from Search Ads 360. If you create a draft in Google Ads, changes you make to the draft do not appear in Search Ads 360 until you apply the draft and sync in the change.

If you create an experiment in Google Ads and sync your account with Search Ads 360, the experiment will appear in Search Ads 360 as a new campaign.

Keep in mind:
  • In Search Ads 360, experiments look like regular campaigns.
  • If you've applied a Search Ads 360 bid strategy to the original campaign, the bid strategy will also be applied to the experiment by default.
  • The original campaign and the experiment use a shared budget. If a budget bid strategy is applied to the original campaign, you should remove the budget bid strategy before you create the experiment because budget bid strategies can't manage spend when the budget is shared.

How to A/B test your AMP campaigns using drafts and experiments

  1. Choose a campaign to test.
    • Ideally, the original campaign that you use to create the experiment should be mobile only.  
    • You can’t segment experiment scorecards by device. You can, however, segment the experiment campaign's results by device, but there will be no statistical significance reporting available. If you don’t have mobile-only campaigns, then (upon the experiment’s end) you’ll need to segment your experiment campaign’s results by device to see how AMP specifically affected mobile performance.
  2. Make sure you have created AMP versions of all landing pages in the campaign.
  3. Click Drafts & experiments in the page menu on the left.
  4. Create a draft of that campaign.
  5. Update the draft campaign’s mobile URLs. Be sure to:
    1. Locate each ad and keyword that has a final URL defined. 
    2. Select “Use a different final URL for mobile”, then paste the AMP campaign url.
    3. Make sure the mobile URL points to the AMP page on your domain (for example, https://example.com/page_amp.html) instead of the Google AMP cache (for example, https://www.google.com/amp/s/example.com or https://cdn.ampproject.org/c/s/example.com).
    4. Leave the base campaign’s URLs unchanged.
  6. Launch an experiment from your newly created draft. We recommend that you:
    • Set a 50/50% budget split so that the experiment version of your campaign is eligible 50% of the time, which should let you see results as soon as possible. Note: You can’t change the budget split once the experiment is running.
    • Use cookie-based splits. This means users will see only one version of your campaign, regardless of how many times they search. This can help ensure that other factors don’t impact your results, and may give you more accurate data.
    • If you are using an audience list, make sure it has at least 10,000 users so that you have enough data to analyze.
    • Select an appropriate start and end date to ensure an appropriate duration and increase your likelihood of detecting a significant result.


  • The more mobile clicks your campaign receives, the faster your test will reach a significant result.
  • When creating your draft campaign, only change the final mobile URL— keep everything else the same for the clearest results. 
  • If you make any changes to your original campaign, those changes won’t be reflected in your experiment campaign (and vice versa). For example, let’s say you want to change your bids; you’ll need to change them in the original campaign as well as the experiment campaign, and you should change both at the same time to avoid skewing your experiment results. 
  • Name your campaigns clearly (for example, “Campaign name_AMP” for the draft) to make them easy to find later.
  • We recommend using drafts and experiments only if you are updating all the final mobile URLs in the experiment campaign. If you are only changing the URLs in a single ad group, we recommend moving that ad group into a new campaign, pausing the ad group in the original campaign, and then running an experiment on the new campaign. 

How to A/B test your AMP campaigns using ad variations

When running an AMP ad variation test, you can segment the test results by device to isolate the performance metrics on mobile.

  1. Click Drafts & experiments in the page menu on the left.
  2. Click Ad variations.
  3. Select the campaign you want to run ad variations for. Keep in mind that you won’t be able to create multiple ad variations for the same campaign in the same date range.
  4. To create a new ad variation, click the plus button .
  5. Click Filter ads if you’d like to narrow down your selection.
  6. Click Continue.
  7. Click the drop-down under “Choose the type of variation you wish to apply to your ads,” and then select Find and replace.
  8. In the “Enter text to find” field, enter the mobile URL that you’d like to replace with the updated AMP URL.
  9. Click the drop down box to the right of the URL you just entered, and then select In final mobile URL.
  10. In the “Enter text to replace” field, enter the updated AMP URL.
  11. Click Continue.
  12. Enter a name for your variation.
  13. Select a start and end date for your test.
  14. Choose a traffic split. We recommend using a 50/50 split so that the variation and the original ad will enter the same number of auctions.
  15. When your ad variation has ended, click the Segment button;Segment in the menu bar and then click Device in the segment dropdown.

Related links

Was this helpful?
How can we improve it?

Need more help?

Sign in for additional support options to quickly solve your issue

Clear search
Close search
Google apps
Main menu
Search Help Center