Splitlab home

Book a Demo

Splitlab home

Book a Demo

Splitlab home

Search for articles…

Which type of test should I use?

Which type of test should I use?

Splitlab offers two kinds of test, and each can split traffic in three ways. Here’s how to pick.

1. A/B test: compare versions of one page

Open A/B Test in the sidebar. You create one or more variations of a page and Splitlab compares each against the original.

Use it when you have a clear idea to try: a new headline, a different offer, a shorter form, a new layout. It’s the simplest test to read and usually reaches a result fastest, because every visitor counts towards just a handful of versions.

2. Multivariate test: find the best combination

Open Multivariate Test in the sidebar. You change several sections of the same page at once, each with its own options, and Splitlab tests the combinations to find the one that performs best. Reports also show how each option did across every combination that contains it.

Use it when you want to know how several changes work together, for example a headline, a hero image and a button at the same time. Because traffic is spread across more combinations, a multivariate test needs more visitors than an A/B test to reach a confident result.

3. Choose how traffic is split

Both kinds of test let you choose a traffic split:

  • Equal splits visitors evenly across all variations. It’s the cleanest choice when you want a statistically clear answer.

  • Custom lets you set each variation’s share yourself, for example to send only a small share of visitors to a risky change.

  • Multi Arm Bandit moves traffic toward the variation that is performing better while the test runs, so you lose fewer conversions to a weaker version. The trade-off is weaker statistical confidence in the final result.

Separately, Traffic allocation controls what share of your visitors enter the test at all. At 100% everyone is placed in a variation; lower it to expose the test to only some of your traffic.

Read more in How the multi-armed bandit works.

4. A quick rule of thumb

  • One idea, one page: A/B test with an Equal split.

  • Several changes on one page and plenty of traffic: multivariate test.

  • Short promotion where earning more during the test matters most: use a Multi Arm Bandit split.

5. Still stuck?

Still unsure which setup suits your page? Get in touch through our contact page or email team@splitlab.io and tell us which page or integration you’re working on.

Still stuck?

Email team@splitlab.io or send us a message.

Contact us


Start optimizing for revenue today not just conversion rates.

Run experiments, track revenue per variation, and scale what actually makes you money.

We’ll increase your conversion rate by 30% or you don’t pay. Terms apply.


Start optimizing for revenue today not just conversion rates.

Run experiments, track revenue per variation, and scale what actually makes you money.

We’ll increase your conversion rate by 30% or you don’t pay. Terms apply.


Start optimizing for revenue today not just conversion rates.

Run experiments, track revenue per variation, and scale what actually makes you money.

We’ll increase your conversion rate by 30% or you don’t pay. Terms apply.