A/B test — Split testing two versions

Showing two versions to different people to see which performs better.

What it means

An A/B test is a simple experiment. You create two versions of something, version A and version B, and show each one to a different group of people. Then you measure which version gets better results.

You can test almost anything: headlines, email subject lines, button colours or images.

How it works

• Decide what you want to improve, such as clicks or sign-ups.
• Change only one thing between version A and version B.
• Split your audience randomly so each group is similar.
• Run the test long enough to collect a fair amount of data.
• Compare the results and choose the winner.

The most common measure is conversion rate:

Conversion rate = (Conversions ÷ Visitors) × 100

Tools like Meta Ads, Google Ads and many email platforms have built-in A/B testing features.

A simple example

A coaching institute in Hyderabad wants more students to book a free demo class. It creates two versions of its landing page:

• Version A: the button says "Submit"
• Version B: the button says "Book my free demo class"

Each page gets 1,000 visitors. Version A gets 30 bookings (3%) and version B gets 50 bookings (5%). Version B wins.

Why it matters

A/B testing replaces opinions with evidence. Instead of arguing about which headline "feels" better, you let real users decide. Small improvements, tested one after another, can add up to better results from the same budget.

Beginner tips

• Test one change at a time. If you change the headline and the image together, you will not know which one made the difference.
• Do not stop the test too early. A few hundred visitors on day one can give misleading results.
• Start with things that can have a big impact, like headlines, offers and CTAs.
• Write down what you tested and what you learnt.

Related: Conversion rate, Landing page, CTA