Roast My Site
Back to Blog
CRO Tips
Apr 2, 20264 min read

A/B Testing for Beginners

EP
Enzo P.
Founder, Roast My Site

Gut feelings are expensive. Every time you change your website based on personal preference instead of data, you're gambling with your conversion rate. A/B testing removes the guesswork entirely.

What Is A/B Testing?

You create two versions of a page element — version A (original) and version B (variant). Half your visitors see A, half see B. After enough traffic, you measure which version produced more conversions. The winner becomes your new default.

How It Works in Practice

1

Identify a page with significant traffic and a low conversion rate

2

Form a hypothesis: "Changing the CTA from 'Learn More' to 'Start Free Trial' will increase signups"

3

Create the variant (version B)

4

Split traffic 50/50 and run the test

5

Wait for statistical significance (usually 95%+)

6

Implement the winner

What to Test First

Headlines

Your H1 has the biggest impact of any single element. Test benefit-led vs. feature-led, question format vs. statement, short vs. long.

CTA Button Text

"Get Started" vs. "Start Free Trial" vs. "Try It Free" — seemingly minor differences can shift conversion rates by 20-30%.

CTA Button Color

See our guide on color psychology for websites for the full breakdown of what works and why.

Images

Hero image with person vs. product screenshot vs. abstract graphic. Test one change at a time.

Free Tools

  • Google Optimize (now replaced by GA4 experiments): Free and integrates directly with Google Analytics
  • VWO (Visual Website Optimizer): More powerful, has a free tier for low-traffic sites
  • Hotjar: Not an A/B testing tool per se, but its heatmaps help you know what to test
  • Mistakes to Avoid

  • Testing too many things at once (you won't know what caused the change)
  • Stopping the test too early because one version looks like it's winning
  • Running tests during unusual traffic periods (holidays, viral moments)
  • Changing the test while it's running
  • Minimum Test Duration

    Never stop a test before 2 weeks, regardless of early results. You need at least 100 conversions per variant to draw meaningful conclusions. Less traffic? Consider testing bigger changes — the signal needs to overpower the noise.

    Share

    Enjoyed this article?

    Get more brutal website advice every week.

    Related Articles