How Split Testing Works in CartFlows: A Beginner’s Guide

Split testing, also called A/B testing, lets you show two or more versions of a funnel step to your visitors and see which one brings in more sales. Instead of guessing whether a new headline, layout or offer works better, you let real visitors decide.

This guide explains what happens behind a split test in plain words: how visitors are shared between versions, which link to share, how results are counted, and how to end a test the right way. For the setup steps, see How to Use A/B Split Testing in CartFlows.

How Split Testing Works

Every split test starts with one of your funnel steps. CartFlows calls your existing step the Control and each new version a Variation. You change whatever you want to test on the variation, and CartFlows sends visitors to the versions for you.

• A new visitor arrives at the step, and CartFlows picks a version for them based on the traffic share you set.
• CartFlows remembers that choice, so the same visitor keeps seeing the same version for the next 30 days.
• Views, conversions and revenue are tracked separately for each version, so you can compare them side by side.

Your customers only ever see one version. They never know a test is running.

How Visitors Are Shared Between Versions

By default, traffic is split evenly. With two versions, each one gets about half of your visitors. You can change this from the Split Test Settings icon next to Start Split Test, for example to send 80% of visitors to your proven step and only 20% to a riskier new idea.

Because each visitor is assigned at random, the split evens out as more people arrive. The numbers become more reliable as more visitors come in, so give the test time before choosing a winner.

Which Link to Share

Keep sharing the same link you already use for the step, the link of the Control version. One link covers every variation. When a visitor opens it, CartFlows takes them to the version they were assigned.

By default, visitors who are sent to a variation see that variation’s own address in their browser. To keep one address for every version, see Using One URL for Every A/B Test Variation.

Checking the Test Yourself

When you’re logged in as an admin, CartFlows shows you the exact step you open, so editing each version is easy. Your own visits are kept out of the test.

To see the test the way a customer does, open the step’s link in a private or incognito window. Each new private window may show a different version, just like a new visitor.

How Results Are Counted

Once you click Start Split Test, the label changes to Stop Split Test and each version shows its own numbers in the funnel:

• Views — how many times visitors opened that version.
• Conversions — how many visitors completed that version, for example by placing an order on a checkout step or moving on from a landing page.
• Revenue — the sales made through that version. Revenue appears on steps where a purchase happens, such as checkout and upsell steps.

The conversion rate tells you more than the raw totals. A version with fewer views but a higher share of conversions may be the stronger one.

Ending a Test the Right Way

When one version is clearly ahead, you declare it the winner. This ends the test and makes the winning version the step all visitors see from now on. For the steps, see Ending a Split Test.

CartFlows keeps the other versions for you in Archived Steps under the step, so you can bring one back for a new test later. Anyone who visits an archived version’s old link is taken to the winner automatically.

Frequently Asked Questions