How to Run A/B Testing on a Website

Andrew Chornyy - 001
Andrew Chornyy

CEO Plerdy — expert in SEO&CRO with over 14 years of experience.

Improving user engagement and conversion rates on a website depends critically on A/B testing there. It’s a way to evaluate two webpage variants to find which one works better. This article will walk you through the processes to execute A/B testing successfully, so guaranteeing that you base your decisions on data to improve the performance of your website.

Step 1: Create Your Hypothesis

  • Find the change: On your page, decide what you wish to modify and the reasons behind it. Users might not be clicking a button on the main page, for example.
  • Goals of the Change: Try to see how tweaks in color or size could affect user behavior.

Step 2: Choose the test URL

  • Decide which page is appropriate. The page where you intend to perform the test should show enough traffic over a two-week period or another suitable length.

Step 3: Define Your Goal

  • Target Page: Choose a page, say a “Thank You” page, that the modification will influence.
  • Site Activity: On the other hand, the objective can be a particular site event—a button click.

Setting Up A/B Testing in Plerdy

How to Run A/B Testing on a Website 02

  • Access A/B Testing: If you have a Plerdy account, click the blue “Create a test” button after visiting Conversions > A/B Testing.
  • Test Name: Enter a name for your test.
  • Test URL: Add the URL where the test will run.
  • Tracking Code: Go to “Settings,” copy the main and extra tracking codes, and add them to the necessary pages including the goal page should the “A/B testing script” not yet have additions. Clear the site cache then confirm the tracking code installation.

Test Variants and Settings

  • End Date: Set when you plan to end the test. Alternatively, manually stop the test after 2-3 weeks if enough data is collected.
  • Audience: Add rules if necessary, including countries and devices.
  • Goals: Add the exact URL with https:// or part of the URL for tracking. Usually, this is the final page affected by Variant B.
  • Events: Add a goal as an event, following instructions for class or ID. Note that event data aligns with the page view limit for heatmaps. Additional limits can be purchased if needed.
  • Sending Events to GA4: Select the checkbox to send event IDs to GA4.
  • Description: It’s advisable to add a note about what you changed and the test’s goal, to remember in 2-3 weeks.

Editing Variant B

How to Run A/B Testing on a Website 03

  • Editing Page: Upon opening the Variant B editing panel, select the checkboxes for “Select element” and “Interactive Mode.”
  • Make Changes: Apply 1-3 changes to an element such as color, size, or hiding it. You can also edit HTML content like links or images.
  • Save Changes: Save individual element changes, and use the “Save all changes” button to save all test modifications.

Launching the Test

  • Review Settings: Open the test settings page in a new tab and refresh.
  • Start the Test: Click the “Start test” button. You can also view all changes made to the website page by selecting “Show all changes.”

Analyzing A/B Testing Results

How to Run A/B Testing on a Website 04

  • Session Distribution: Every variant gets almost 50/50% user sessions.
  • Key Metrics: Search the Total Sessions tab’s “Improvement” column for the winning version.
  • Device Analysis: Review which device most affected the winning variant.
  • Traffic Analysis: Find out which traffic channel worked best for you.

Interpreting Negative Values for Variant B

How to Run A/B Testing on a Website 05

Should Variant B exhibit a negative value in the improvement metric, Variant A might outperform Variant B. Here, you have two choices:

  • Wait a Few More Days: Sometimes low traffic or initial user behavior modification could cause first results to not show the actual performance. A few more days of waiting will supply more information necessary for a certain conclusion.
  • Take Variant B into account. Unsuccessful: One could be safe to assume that Variant B is underperforming relative to Variant A if the negative trend keeps regularly throughout a notable duration. Under such circumstances, it is advisable to examine the aspects of Variant B that might be generating the performance drop and take into account either changing or eliminating the modifications done in this variant.

Tips for Checking Variant B

  • Multiple Browsers: Check Variant B using many browsers since you cannot see both variations in the same browser session in half an hour.

Final Thought

One effective instrument for improving a website is A/B testing. These guidelines will help you to make wise judgments grounded on user preferences and behavior. Recall that effective A/B testing depends on ongoing learning and adaptation. delighted testing.

Was this helpful?