How to AB test landing pages in Google Ads, step by step

How to AB test landing pages in Google Ads: the method

To A/B test a landing page in Google Ads, run two versions of the page at one web address, change one thing between them, set the traffic split before launch and wait until enough visitors have seen both to show a clear winner. An experiment tests variations of one page; the traffic split is the share of new visitors sent to each variation.

Google Ads has its own experiments too. A custom experiment makes a trial copy of a campaign and splits traffic between the two, and Google lists landing pages among the things it is typically used to test. Google's experiment needs no builder but splits a campaign; a builder's test splits the visitors of one page. Google Optimize, Google's old page-testing tool, ended on September 30, 2023.

Set up the test: one page, one change, one web address

Instapage's help center documents these steps:

1. Publish the page first. Experiments can only be created for pages already published to a live URL.
2. Open the Optimize tab, click Experiment in the right sidebar, then Create Experiment.
3. Name the experiment and write the hypothesis, the change you expect and why.
4. Set the traffic split. It need not be even but must add up to 100%, and no variation gets 0%.
5. Start the test and leave it alone. The split cannot be changed after the start, and editing a running experiment makes it unreliable.
6. End the test by choosing a winning variation, which then gets the live address.

Returning visitors keep the version they first saw, and the experiment view counts only unique visitors and conversions.

How many variations and how long to run

Instapage recommends at most 10 variations and warns that the risk of a false positive rises with every extra variation. A small budget points to few: each variation splits the same visitors further. Instapage says to run the test until enough viewers have seen it to show a clear winner, and it gives no number.

None of the builders' pages gives a minimum number of visitors for a reliable test. Google's only figure is for its own custom experiments: more than 100 daily conversions on the base campaign as an example, and at least 4 to 6 weeks when results are unclear. Compare that with your account's conversions a day. What to change on the page is in landing page optimization for Google Ads.

Testing on a small ad budget

Instapage offers a 14-day trial of Create or Optimize, and testing starts on Optimize. A credit card is required to start, and it is charged when the trial ends unless you cancel. The trial also ends at 2,500 unique monthly visitors (different people, counted again after 3 days), when the paid plan starts.

A test needs a published page. Instapage's pages do not agree on whether a trial account can publish live pages, so confirm that in the trial before sending ad traffic.

Where A/B testing starts in each builder

Instapage puts server-side A/B testing on Optimize, $199 a month, billed monthly as read on October 18, 2026, with 30,000 unique monthly visitors, and not on Create. Server-side means the server shows each visitor one version of the page.

Instapage pricing lists the other plan limits, and Instapage vs Unbounce compares the two plans side by side.