A/B testing: evaluate changes through a controlled experiment
A new page version may seem easier to use, but its impact needs to be measured. A/B testing lets you compare variants using a defined metric and gather evidence for a decision.
We help formulate a hypothesis, plan an experiment and analyse the result. Before launch, we check whether there is enough traffic and enough target events for the chosen objective.
What you can test
- Offer presentation and key arguments.
- Landing page structure.
- Forms and individual steps in a user flow.
- Call-to-action placement and wording.
- Product page or checkout elements.
How we conduct testing
Hypothesis and metrics
We describe what is changing, why we expect an effect and which action we will measure. We define the primary metric and metrics that help detect unintended consequences.
Experiment plan
We estimate the required data volume and potential duration. We agree on the audience, allocation between variants and rules for ending the test.
Implementation and verification
We prepare variants within the agreed scope. Before launch, we check version display, user allocation and tracking accuracy.
Result analysis
We assess the difference between variants and uncertainty in the result. We check data quality and adherence to the experiment plan.
What you receive
A conclusion about the test and a recommendation for the next step: implement the change, retain the current version or test another hypothesis.
A test may not provide sufficient evidence to select a winner. In that case, we explain the data limitations and assess whether further investigation is worthwhile.
Let’s test your hypothesis
We will discuss the change, metric and feasibility of running an experiment on your website.