Skip to content

Conversion Rate Optimisation (CRO)

A/B testing

Make website decisions with data, not opinions.

What's included

  • Hypotheses
  • Variations
  • Execution
  • Documented decision

Part of Conversion Rate Optimisation (CRO)

The essentials

Why A/B testing matters

In an A/B test, half the visitors see the current page and the other half see a variation. At the end, the data says which converts better, instead of the decision depending on whoever argues loudest.

We only test what has a clear hypothesis behind it, and only decide when there's enough data.

What we do

What's included

  1. 01

    Hypotheses

    What to test, why and what result we expect.

  2. 02

    Variations

    Copy, layouts and elements changed for each test.

  3. 03

    Execution

    Tests with a duration and sample suited to your traffic.

  4. 04

    Documented decision

    What won, what we learnt and what goes live on the site.

How we work

Step by step

  1. 1

    Hypothesis

    What changes, why and which metric should move.

  2. 2

    Design

    The variation, with copy and design.

  3. 3

    Run

    Until there's enough sample and duration.

  4. 4

    Decide

    Roll out the winner and document what we learnt.

Platforms and tools

What we work with

  • A/B testing tools
  • Google Analytics 4
  • Google Tag Manager

What we report

Every month, compared with the previous period.

  • Conversion rate
  • Average order value
  • Cart and form abandonment
  • Revenue per visitor

Case study · Skincare, e-commerce (Norway)

Flora Organics: +205% orders during Black Week

A Shopify store built from scratch, with website, SEO, social media, ads and measurement working together for a Norwegian skincare brand.

+205%

orders

Black Week 2025 vs 1–22 Nov 2025

More in Conversion Rate Optimisation (CRO)

Other specialities

See everything in Conversion Rate Optimisation (CRO)

FAQ

Before we talk

Do I need a lot of traffic?

It depends on what you want to measure. With little traffic, we prioritise bigger changes or other ways of validating.

Do tests slow the site down?

We set tests up to have minimal impact on speed.

How long does a test run?

Usually a few weeks, to cover full buying cycles. It depends on traffic.

What if neither version wins?

That's an answer too: the change makes no difference and we can move on to the next hypothesis.

Shall we talk about A/B testing?

Tell us the context and we'll come back with the next steps.