By Solo Unicorn · 14 September 2026 · 2 minute read
Before you begin
State the hypothesis, audience, channel, measure and time window before starting. Compare like-for-like results. Avoid changing the message, audience and offer at the same time. Record what happened and what you would do differently. Small samples are uncertain, so look for repeated evidence before making a major commitment. Growth follows learning and delivery capacity together.
Write a narrow hypothesis
An example experiment could ask whether a clearer product photograph helps visitors understand a quotation page. Keep the offer and intended audience consistent while changing the photograph. Decide which behaviour matters: reaching the product details, starting an enquiry or completing a useful request. A page-view increase alone does not establish that the business received better leads. Record the starting conditions before publishing the change.
Plan how to interpret the result
Choose a review period and note anything else that changes, such as a promotion or a different traffic source. Small samples can be noisy. Record the counts behind percentages and resist a strong conclusion from a handful of visitors. Include qualitative feedback from enquiries. If customers repeatedly ask a question the page should answer, that may be a clearer improvement opportunity than a tiny change in a metric.
Keep a decision record
Lean Growth Experiments should end with a practical next step: keep the change, revert it or gather more evidence. Write the reason and the uncertainty. Productised Services gives you a clearer offer to test, while Founder Systems helps the team handle the enquiries that follow. Do not scale an experiment just because its headline number looks positive; check whether the resulting work fits your capacity.
Put the notes to work
Prepare a brief with the free planning tool →