How-to guides

How to Run an AB Test That Actually Improves Your Conversion Rate

AB testing shows which page version converts more visitors. Follow this step-by-step guide to run an AB test, avoid mistakes, and make data-backed decisions.

Published

How to Run an AB Test That Actually Improves Your Conversion Rate. A staircase of dots with 5 steps, one for each step in the guide.

An AB test compares two versions of a page by showing each to a random half of your visitors. The version that gets more people to take the action you care about—like submitting a form or clicking a button—is the one you keep. Here’s how to set up, run, and learn from an AB test without wasting time or traffic.

The core idea of an AB test

An AB test answers a simple question: “Does this change make more people say yes?” You take your current page, change one element on a new version, and send half your traffic to each. Measure a clear action—form submissions, button clicks, or purchases. When one version clearly outperforms the other, you have a winner. If they perform almost the same, keep the original and test something else next.

Most pages don’t convert as well as they could. Across pages published with Leadpages, the median conversion rate is 15.7%. That means half of pages do better than that, but half do worse—with the bottom quarter converting below 4.66%. Small, systematic tests can move a page from the bottom quarter toward the median, or from the median toward the top quarter, where pages convert above 39%.

How to run an AB test in 5 steps

1. Pick the one thing you want to improve

Don’t change your headline, image, button color, and form fields all at once. Change one variable so you’ll know what caused the difference. Common first-test candidates:

  • Headline
  • Call-to-action button text
  • Form length
  • Hero image or background video
  • Social proof (testimonial placement, star rating, logobar)

If you’re unsure what’s holding your page back, run it through a landing page analyzer to see where you’re losing visitors. A headline analyzer can also give you a quick score before you commit to a variation.

2. State your hypothesis clearly

A hypothesis isn’t “The new version will do better.” It’s “If we move the testimonial above the form and keep everything else the same, then more people will see social proof before they decide, which will increase form submissions.”

A good hypothesis has:

  • The specific change you’re making
  • The page element you’re changing
  • The outcome you expect
  • A reason grounded in visitor behavior

Write your hypothesis down. It keeps you honest when you read the results.

3. Create your two page variants

Version A is your control—the current page. Version B is the variation with your one change. Build it in your landing page builder so the rest of the page remains identical. If you use Leadpages, you can clone your page, make the change, and turn on AB testing in a few clicks.

Keep the thank-you page (or whatever happens after the goal action) the same for both variants. You’re measuring the page’s ability to prompt the action, not what comes after.

If you need a starting point for your control or variation, browse a template gallery to find layouts that already perform well in your industry.

4. Split your traffic and let the test run

Send 50% of your visitors to version A and 50% to version B. Make sure the split is random and consistent—most page builders and AB testing tools handle this automatically.

Then leave the test alone. Checking results hourly leads to bad decisions. You need enough visitors and conversions for reliable numbers. As a rule of thumb, stop only when:

  • The test has reached statistical significance (often 95% confidence or higher), and
  • You’ve collected data for at least one full business week, so day-of-week patterns don’t skew things.

Use an AB test significance calculator to see if your numbers are trustworthy before you declare a winner.

5. Act on the outcome

Once the test runs its course, you’ll see one of three results:

  • Version B won. Turn off the A variant and make B the new permanent page. Then start your next test on the new page.
  • No clear winner. The difference was too small to matter. Keep the original page and test a different element—maybe a bolder change—next time.
  • Version A won. Your change hurt conversions. Revert to the control, and learn from the insight. Now you know what your audience doesn’t respond to.

Resist the urge to tweak a borderline result into a “winner.” If the difference isn’t significant, it’s noise.

AB testing problems that waste your time (and how to avoid them)

Not enough traffic

If your page only gets a couple hundred visitors a month, a standard AB test can take months to finish. In that case, test only high‑impact changes that could produce a big lift—like a completely different headline or a new value proposition—so you can spot a result with fewer visitors. Or focus first on driving more traffic before you split-test.

Testing minor tweaks that don’t move the needle

Changing a button from green to blue won’t likely change your conversion rate. Test things that influence a visitor’s decision: the offer, the headline, the reason to act now, or the amount of social proof. If you’re not sure what moves the needle, look at pages in the top quarter of conversion rates—across our sample, those pages convert above 39%—and note what they do differently.

Ending the test too early

Impatience is the most common AB testing mistake. A result that looks like a winner after two days might flip after a full week. Wait until your test reaches significance. If you use Leadpages’ built‑in AB testing, the system shows you the statistical confidence level so you’re not guessing.

Running tests that overlap

If you run an AB test on a page while also changing ad targeting or sending a different email subject line, you won’t know what drove the change. Keep your AB test isolated—change only the page element, and keep other traffic sources consistent.

Ignoring what you learned

Treat every finished test as an asset, even the “losers.” Keep a simple log: what you tested, your hypothesis, the result, and what you’d try next. Over time, you’ll see what your audience responds to and stop testing things that don’t matter.

Turn your test result into more leads

Your AB test gave you one winner on one page. Now make that learning work harder for you.

Scale what worked

If a headline style, testimonial placement, or button text lifted conversions, roll that same pattern onto other pages. Similarly, if something consistently fails, stop doing it across your site. The top tenth of pages in our dataset convert above 62.4%, often because those marketers apply the patterns that win for their audience.

Test the next thing

The winning page is now your new control. What’s the next most promising change? Look at your page’s weakest section—maybe the area where visitors scroll away—and write your next hypothesis. A free call-to-action generator can give you fresh button copy ideas if you’re stuck.

Optimize the full funnel

An AB test that lifts your landing page conversion rate is great, but what happens after the click? If your follow‑up email sequence or sales call process leaks leads, the page improvement won’t show up in revenue. After your page tests, look at the next step in the flow and start improving that.

Where the numbers come from

The conversion rate figures above come from an analysis of more than 13,000 pages published with Leadpages. All data is aggregated and anonymized, current as of 2026-11-16. The median conversion rate across all pages is 15.7%. The bottom quarter convert below 4.66%, the top quarter convert above 39%, and the top tenth convert above 62.4%. Those benchmarks show what’s possible when you test and improve systematically.

If you don’t yet have a tool to run AB tests on your landing pages, start a free trial and publish your first test today.

FAQ

Questions, answered

AB testing is the process of comparing two versions of a web page by splitting traffic between them. You measure which version gets more visitors to complete a goal—like a form submission or a purchase—and let the data tell you which one to keep.

Run the test until you reach a statistically significant result and have at least one full business week of data. The exact time depends on your traffic volume and the size of the conversion difference, but stopping early often leads to false conclusions.

Statistical significance tells you the result is likely real, not random chance. Most marketers aim for 95% confidence or higher. You can use a free AB test significance calculator to check your numbers without doing the math yourself.

Start with the element that most directly affects a visitor’s decision: the headline, the main offer, or the call-to-action. These tend to create bigger lifts than small design tweaks, so you’ll see results faster.

Yes, you can run an AB/n test with three or more versions, but you’ll need more traffic to reach significance. If your traffic is moderate, stick to two versions per test and cycle through variations one at a time.

Ready to build?

Put it to work on your own page

Build, test and improve landing pages in one place.