What is A/B Testing

A/B testing in Google Ads is a marketing tool where two different elements in an ad, like text, call-to-action, images, or links, are compared to find the most effective ones.

These tests are conducted to understand which ad elements work best for our audience and, as a result, increase conversions with the same advertising budget.

Success Criteria for A/B Experiments

  1. Formulating a hypothesis. For instance, with a low CTR, the assumption could be: “Ad CTR can be increased by inserting a call-to-action in one of the headlines and placing it second”.
  2. To eliminate the possibility of external factors influencing the results, traffic between the original and experimental ad versions should be equally split.
  3. For valid results, only one hypothesis should be tested within a single experiment.
  4. Avoid making any changes or concluding the test until statistically significant results are obtained. Each ad variant should receive at least 100 clicks. The more, the better.
  5. Identifying the most effective variant and applying it.
  6. Formulating new hypotheses and conducting A/B tests regularly.

The best and simplest way to conduct A/B testing in Google Ads is to use Google’s special internal functionality called Experiments.

In the article, we will share a case study on how we tested two landing pages for our client and the conclusions we reached. Prepare to give your attention for 7 minutes.


Our client is a renowned German manufacturer of horse saddles. Since 1999, they have been developing innovative saddles, handcrafted and tailored to meet any rider’s needs.

We have been working with the client since August 2022. During our collaboration, they expressed a desire to increase the form submission rate from their website. Previously, advertising led directly to the site, which we continually optimized and achieved good results.

However, there’s always room for improvement, so after consulting with us, we decided to develop and conduct an A/B test between the standard site and a newly created lead-generating landing page.

The small landing page featured a lead magnet in the form of a form where people could submit their information and get a chance to download a price list for free.

A/B testing in Google Ads without Google Optimize

Implementation Process

Google has several types of experiments.

Types of experiments for AB tests in Google

The most popular types of experiments in Google Ads include

Custom experiment. Used to test different bidding strategies, keyword match types, audiences, etc., within a single campaign.

Optimize Text ads. Used to test various changes in the ads themselves for one or multiple campaigns.

How to Conduct an A/B Experiment in Google Ads

1. Log in to your account.
2. Create a duplicate of the campaign you want to test in an A/B test.
3. Click “Drafts & Experiments” → “Campaign Experiments” → + button.
4. Select your duplicate, name the experiment, and choose a start date.
5. Set the advertising budget. Choose the percentage of the campaign budget participating in the experiment, for example, 50%.
6. Save your experiment and wait for the results.
That’s it. Google Ads has its own calculator that does all the calculations and shows which variant won.

Julia Sotnikova, CVO marketing.link

Expert Comment

In all A/B testing, a common mistake is having too small a sample size. The sample must be representative and accurate. Applying experimental data with a too-small sample size to the entire project can be hazardous for the business.

Here’s a list of the most common mistakes in A/B experiments:

  • Insufficient sample size.
  • Testing without a clear hypothesis: testing for the sake of testing.
  • Testing minor and major changes, such as testing similar images or, conversely, testing many variables at once: images, text, and more.
  • Ending the experiment too early and getting discouraged at the first failure.
  • Not documenting test results.
  • Ignoring intuition or, conversely, placing too much trust in feelings when experiment results are adjusted to fit the experimenter’s wishes.

Julia Sotnikova, CVO marketing.link

Optimize Text ads Experiment 

Let’s discuss the process of creating a Optimize Text ads experiment. At the campaign level, click the Experiments button at the bottom.

Experiment Optimization of text ads

Click the plus sign.

The process of creating an experiment Optimization of text ads

Among all the options presented, select Optimize Text ads and click Continue.

the process of creating an experiment in Google Ads

Select the campaigns in which to conduct the test.

How to conduct an AV experiment with Google image ads

Next, filter the ads by a parameter. If you’re testing a URL, specify the primary URL you will compare with another.

How to conduct AB testing in Google step by step

Then create an experimental ad version. Specify what changes to make for the test in the ads filtered previously.

Stages of split testing in Google Ads

Enter the experiment name, its start and end date, and what percentage of traffic to send to the changed ads. We recommend specifying 50%.

How to run AB experiment in Google Ads

The experiment is ready. Do not make any further changes to it. Allow the system some time to work quietly, testing different ad versions. The more, the better, for greater reliability and statistical significance of the result.

After accumulating data, you can evaluate the results, apply changes to the base campaign, or discard them.

It’s worth noting that the Google Optimize tool for experiments ceased operation in September 2023. All experiments can now be conducted directly in the Google interface.

What results were obtained

We tested the new landing page for a month. The experiment showed that a full-fledged website is more conversion-effective than a lead-generating landing page.

With a comparatively equal amount of traffic received, ads leading to the test site got 59% fewer conversions than those leading to the full website, and the cost of acquiring one conversion was more than 3 times higher.

What results were obtained as a result of AV testing


Never rely solely on personal opinion, even with significant experience in advertising. Sometimes the results can be unexpected, as in our case.

Therefore, always verify your hypotheses through experimentation and make decisions based on real numbers, not subjective assumptions.

Tags of this page:
Leave a comment

Your email address will not be published. Required fields are marked *