What Is Split Testing (A/B Testing) and How Do You Do It?

Split testing — also called A/B testing — is the practice of showing two different versions of a web page (or a specific element on a page) to different visitors at the same time, then measuring which version performs better against a defined goal. Version A is typically your existing page (the control), and version B is the variation with one change — a different headline, a new call-to-action colour, or a rearranged layout.

The principle is straightforward: instead of guessing which version of a page will convert better, you let real visitor behaviour decide. Over time and across enough visitors, the data reveals a statistically significant winner. This removes opinion from design and marketing decisions, replacing gut feel with evidence.

What Can You Split Test?

Almost any element on a page can be tested, but the highest-impact tests focus on elements that directly influence conversion decisions. Headlines are consistently the most impactful — a different headline can shift conversion rates dramatically because it’s the first thing visitors read. Call-to-action button text and colour are close seconds. “Get a Free Quote” versus “Request Your Quote” might sound like a small difference, but can produce measurably different click rates.

Other high-value test candidates include hero images (lifestyle vs product-focused photography), the presence or absence of social proof (testimonials, review stars, trust badges), form length (asking for fewer fields typically increases submissions), and pricing presentation (monthly vs annual toggle, price anchoring). On e-commerce product pages, testing the main product image and the layout of the add-to-cart section can yield significant revenue gains.

How to Run an A/B Test

The most widely used free tool for split testing is Google Optimize — though as of 2023, Google has wound this down in favour of integration with Google Analytics 4 (GA4) experiments. Third-party tools like VWO, Optimizely, and Convert.com offer more sophisticated testing capabilities. Shopify has built-in A/B testing for themes, and many email marketing platforms include split testing for subject lines and send times.

To run a valid test, you need enough traffic to reach statistical significance within a reasonable time frame. A page receiving 50 visitors a month will take many months to produce reliable results on most tests. As a rough guide, aim for at least 100 conversions per variation before drawing conclusions. Set your test up for a minimum of two weeks to account for day-of-week effects, and avoid running tests during unusual periods (major promotions, holidays) that might skew results.

Reading and Acting on Results

Most A/B testing tools report a “statistical confidence” percentage. Aim for 95% confidence before declaring a winner — this means there’s only a 5% chance the observed difference is due to random variation rather than a genuine effect of the change. Stopping a test too early (a common mistake) can lead you to implement changes that are actually neutral or slightly negative.

When a test produces a winner, implement the change permanently and document what you tested and what you learned. Over time, a library of test results becomes an invaluable guide to what resonates with your specific audience. After implementing a winner, wait a few weeks before running the next test on the same page element, to allow your analytics baseline to stabilise before starting the next experiment.