A useful marketing experiment starts with a decision, not an edit. Before changing a page, email or offer description, write down what you want to learn, what you will change and what evidence would justify keeping it. A one-page plan makes those choices visible without turning a small task into a research project. Its purpose is to help you decide what to do next—not to manufacture a winning result.
Start with the decision you are avoiding
“Improve my website” is too broad to test. “Should I explain the deliverable before asking visitors to enquire?” points towards a specific decision. Choose something you can act on yourself, reverse if necessary and evaluate with evidence you can realistically collect. If answering the question would require redesigning the whole site or attracting a much larger audience, narrow it first.
Example: a solo designer offers a fixed-scope brand audit. The service page describes the process, but the designer wonders whether visitors understand what they will receive. A manageable question is: “Does placing a short deliverables list above the enquiry button help prospective clients understand the audit before contacting me?” This asks about one communication problem rather than whether an entire page is “better”.
Use six fields to keep the plan on one page
Write the plan before making the change. Keep each field to a few sentences; extra detail is useful only when it prevents confusion later. The following template works for a landing page, newsletter invitation or service description.
- Decision and question: What will you decide afterwards, and what specific uncertainty stands in the way?
- Hypothesis: What do you expect to happen, and why might this change help the intended reader?
- Change and constants: Name the single meaningful edit, plus the offer, audience and other elements you intend to leave alone.
- Evidence: Choose one primary signal linked to the question, with limited supporting feedback.
- Schedule and safeguards: Set a start date, review date and conditions that would require an early stop.
- Decision rule: Describe what would support keeping, reversing or revising the change—and what would remain inconclusive.
Match the evidence to the question
In the designer example, the hypothesis might be that a visible deliverables list reduces uncertainty about what the audit includes. Direct feedback from relevant prospective clients could address that understanding question. Enquiry counts alone cannot show whether visitors understood the offer; they may change because of traffic quality, timing or unrelated events.
Example trial: show the existing and revised page separately to people resembling the intended audience. Ask neutral questions such as “What would you receive?” and “What is still ambiguous?” Avoid prompting them to praise the new list. Record recurring misunderstandings, not just preferences. This is a small comprehension check, not proof that the revised page will generate more enquiries.
If the real question concerns enquiry behaviour, the plan needs behavioural evidence instead. Check that enquiry tracking works and that you can relate enquiries to page visits. A before-and-after comparison is simpler to organise, but changes in audience or season can muddy the result. Randomly splitting comparable traffic can reduce some of that ambiguity, yet low traffic may still leave the answer uncertain.
Choose a review date without forcing a verdict
Set a date that fits the normal rhythm of your marketing activity and your available time. Treat it as a checkpoint, not a promise that enough evidence will exist. Write down any planned promotion, holiday or offer change that could affect interpretation. For a numerical comparison, a calendar deadline alone does not establish that the sample is adequate.
Define early-stop conditions separately from success. A broken enquiry form or inaccurate service description needs immediate attention; a quiet afternoon does not justify declaring failure. When evidence is thin, choose deliberately between extending the experiment, using a more direct research method or accepting that this question is not currently worth the effort.
Check the plan, then make the smallest useful change
- Can you answer the question without changing the offer, audience and message at the same time?
- Have you checked the starting page, saved a copy and confirmed that forms or tracking work?
- Does the chosen evidence address the question, rather than merely provide an easy number?
- Have you allowed for an inconclusive result and written down the next decision?
Draft those six fields for one unresolved marketing decision. If the plan feels crowded, reduce the question rather than shrinking the writing. After the review, add a short note recording what you observed, what remains uncertain and what you will do next. That makes even an inconclusive experiment useful: it leaves a more focused basis for the next choice.