What is an example of an A/B test? 10 real examples and their results

A common example of an A/B test: an e-commerce store shows half of its mobile shoppers a standard add-to-cart button, and the other half a sticky version of the same button that remains visible as they scroll. The winning variation, of course, is the version that gets more shoppers to add items to their cart.

It’s a typical test, but one with real potential: when No Two Houses ran this test, their add-to-cart rate increased by 58%.

That’s the main idea behind A/B testing: split your traffic at random between two versions of the same experience. Then, measure a primary metric for both versions or variations, and keep the version that your data supports.

Below, we’re sharing ten real A/B test examples we’ve seen from teams experimenting using Kameleoon. We’ll share what changed, what happened, and you could replicate the test yourself using Kameleoon’s Prompt-Based Experimentation (PBX).

What is an A/B test?

An A/B test is a way to compare two or more versions of a page or experience to see which one works better. You can test something as simple as the color of some text or as complex as an onboarding flow. By showing half of your visitors one version of the page and the other half another, you can compare the performance of each to determine if one is more beneficial for conversions, user experience, or any other metric. 

A/B tests can also be used to evaluate personalized experiences, involve multiple variables, and apply different statistical techniques to help interpret results. To learn more about what an A/B test is, you can read this article.

What can you test with an A/B test?

There are many things you can test with an A/B test based on two broad categories: client-side and server-side testing. Client-side A/B tests typically run via JavaScript, which means they affect front-end variables: text, colors, shapes, forms fields, and popups, for example. 

Server-side tests take place on the developer’s side, controlled through backend code, and can therefore be more complex: pricing algorithms, software features, and checkout queues. 

You can A/B test nearly any aspect of your digital experience with the right tools, teams, and process.

10 real A/B test examples

Each of the following examples comes from one of Kameleoon's customer stories, allowing us to highlight the challenges, hypotheses, and wins that each example could generate. Several were even built using PBX.

Simple A/B test examples: edit what’s already there

These two tests are the kind of straightforward yet impactful experiments most marketing specialists can run. Both variations are largely the same, testing simple copy or layout changes only. 

{{blue-block-1}}

‍

A/B test examples for adding a simple element

These tests step up the difficulty slightly, requiring basic Javascript or CSS to build new elements on otherwise unchanged pages. An e-commerce specialist would have no trouble building these out via pre-made widgets; a more senior marketer could perhaps code them by hand. 

{{pink-block-1}}

‍

Challenging A/B test examples: building new behaviors

These tests involve interactive components and would require across-device testing and custom Javascript, in some cases across thousands of webpages. These require some developer input and may require a few days to create. 

{{blue-block-2}}

‍

Complex A/B test examples: change the flow or data

These tests pull content from other systems or webpages or change how multi-step flows work. This is where testing can get very complicated, and both examples would require a developer sprint to make work.

{{pink-block-2}}

{{quote}}

An example where the original won

A losing A/B test still returns a useful result by stopping you from shipping a change that might have cost conversions.

The Alberta Motor Association, for example, tested a new call-to-action button against its original. When the original won, the team knew not to release the change. 

The same team then tested a change to its membership comparison page and found a variation that improved conversions by 10%. 

So one test found a win and another prevented a loss. In both examples, the tests were worth the time committed to them. 

{{quote1}}

{{cta-block}}

‍

From example to live test with PBX

Every A/B test example you’ve read about here started with an idea that can be expressed in a plain-language sentence. With Prompt-Based Experimentation (PBX), that sentence builds the whole test, including each step from idea to the conclusion of the test:

• PBX Ideate analyzes your pages and proposes test ideas, ranked by a model trained on over 20,000 experiments.
• PBX Build turns a prompt or Figma frame into a production-ready variation, using your full site as context.
• PBX Configure sets up audiences, goals, and traffic allocation through the same chat window as PBX Build.
• PBX Ship (MCP server) lets developers pull winning variations into their IDE and ship them behind feature flags.

Want to try it yourself? Choose one example from this list, describe how it might work on your own site in a simple sentence, and try PBX for free to build the experiment in minutes.