G
GrowWebPro
Analytics & Data

What Is A/B Testing and How Does It Work?

A/B testing shows two versions of a page or element to different visitors and measures which one converts better. Here's how to run your first test.

2 min read · Updated 2026-05-10

Short answer

A/B testing (also called split testing) shows version A to 50% of visitors and version B to the other 50%, then measures which drives more conversions. Change one element at a time — headline, button text, image, price — so you know what caused the difference.

A simple example

You want to know if "Get a Free Quote" or "Request a Callback" performs better as a button label.

  • Version A (50% of visitors): button says "Get a Free Quote"
  • Version B (50% of visitors): button says "Request a Callback"

After 500 total visitors, version A generates 30 clicks (6%) and version B generates 18 clicks (3.6%). Version A wins — you make it the default.

What to test first

Headline — the biggest lever on any page. A better headline can double conversions. Test: make a direct benefit claim vs a question vs a how-to statement.

Call to action button — text, colour, placement. "Start my free trial" often beats "Sign up." "Call us" often beats "Contact us."

Hero image or above-the-fold content — does a photo of a person outperform a product photo?

Pricing presentation — does showing monthly price before annual increase or decrease sign-ups?

Tools for A/B testing

Google Optimize shutdown note: Google Optimize was discontinued in 2023. Use one of these alternatives:

  • VWO (vwo.com) — starts at $199/month; full-featured
  • Optimizely (optimizely.com) — enterprise tool, expensive
  • AB Tasty (abtasty.com) — mid-market option
  • Crazy Egg — includes simple A/B testing starting at $29/month
  • Your email platform — most email tools (Mailchimp, GetResponse) have built-in A/B testing for subject lines and content

For most small businesses, start with A/B testing your email subject lines (free in most tools) before investing in a page testing platform.

How long to run a test

Run until you have statistical significance — usually when each version has at least 100 conversions. Don't stop after 2 days because one version looks like it's winning — small samples are misleading.

A rough guide: if you get 50 conversions per week, run the test for at least 4 weeks.

Related questions