A/B Testing
A/B Testing &
Experimentation.
Hypothesis-driven experiments that answer specific questions about your funnel — not gut-feel guesses dressed up as science.
How it works
Most A/B testing programmes are run wrong.
Running a test because something "feels wrong" is not experimentation — it's guessing with a dashboard attached. Real A/B testing starts with user behaviour data, forms a specific belief about why conversion is low, and designs a test to confirm or refute that belief.
The output of a well-run testing programme is not just a higher conversion rate. It's an accumulating body of knowledge about your customers — what language they use, what they fear, what motivates them — that makes every future decision faster and more confident.
How I run tests
Every test has a hypothesis
A/B tests without hypotheses are coin flips with extra steps. Every test I run starts with a specific belief about user behaviour and what changing it will do to revenue.
Statistical significance matters
Running tests too short, with too little traffic, or declaring winners too early is how companies fool themselves. I run tests to proper significance and document sample size requirements upfront.
Losers are as valuable as winners
A test that shows no improvement tells you something important about your customers. The goal is learning, not just winning — learning compounds.
One variable at a time
Multivariate tests have their place, but most businesses need to know *what* moved the needle. Controlled tests, properly isolated, give you that answer.
What gets tested
Headline testing
The highest-leverage test on most pages.
CTA copy & placement
Colour, copy, position, and surrounding context.
Social proof placement
Reviews, logos, and trust signals positioned for maximum impact.
Form optimisation
Field count, order, labels, and error handling.
Pricing presentation
How price is framed, anchored, and contextualised.
Page layout & hierarchy
What users see first and what they see next.