Start with the decision, not the method
Business questions often arrive in broad terms: “Why are sales slowing?” or “Should we change our offer?” These are reasonable starting points, but they are not yet research plans. Before choosing a survey or scheduling interviews, identify the decision someone needs to make. Is the team deciding which feature to improve, which audience to serve, or whether to revise a message? A clear decision gives the research a purpose.
Write the decision in a sentence that names the options. For example: “We need to decide whether to simplify our sign-up process or explain its benefits more clearly.” Then note who will make that decision and when they need the findings. If the team cannot name the options yet, the first research task may be exploratory: learning how customers describe the problem before testing a specific solution.
Separate what is known from what is assumed. A drop in sign-ups may be visible in internal records, while the belief that people find the form confusing is a hypothesis. Treating assumptions as questions prevents a plan from becoming an exercise in confirming a preferred explanation.
Turn the decision into answerable research questions
A research question should describe something you can realistically learn from people or existing information. For the sign-up example, useful questions might include: Where do potential customers hesitate? What do they expect to happen after signing up? Which details do they need before they feel comfortable continuing? Each question connects to the decision without presuming the answer.
Prioritize the questions that could change what the team does. It may be interesting to learn how people first heard about the business, but that information may not help choose between simplifying a form and improving its explanation. A short list of essential questions keeps the project manageable and leaves room for unexpected findings.
Define the terms you use. “Customers,” for instance, might mean recent buyers, people who abandoned a purchase, or anyone who has considered the product. Those groups may have different experiences. Specifying the audience now helps prevent a mismatch between the evidence collected and the decision it is meant to inform.
Choose evidence that fits the question
Different methods reveal different things. Existing service messages or website records can help locate recurring issues or points where people leave a process. Interviews can explore how people interpret an offer and why they make certain choices. Observing someone attempt a task can reveal friction they might not recall in an interview. A survey can help compare reported experiences or preferences across a defined group, provided its questions and recruitment are carefully designed.
Choose methods for their fit, not their familiarity. If you need to understand language, expectations, or an unfamiliar problem, begin with open-ended conversations. If you already know the main possibilities and need to compare responses, a structured survey may be more useful. Combining methods can also help: interviews may reveal potential explanations, while a later survey can examine how widely those explanations are reported among the people reached.
Be explicit about limits. A small set of interviews can provide depth but cannot establish how common an experience is. A survey of volunteers may capture useful views without representing every customer. Record who the evidence covers, how it was gathered, and which conclusions it can reasonably support.
Set the scope before collecting anything
A workable plan specifies whom to include, how to reach them, what to ask or observe, and how much time is available. Recruitment deserves particular attention. Speaking only with loyal customers could miss obstacles faced by newcomers; speaking only with people who left could miss what currently works. Select participants according to the decision, rather than simply choosing whoever is easiest to contact.
Keep the plan respectful and proportionate. Explain the purpose of participation, avoid collecting personal details you do not need, and give people a way to decline. Review questions for leading wording, such as “How helpful was our improved sign-up page?” A more neutral question asks what the person expected from the page and what, if anything, was unclear.
Also decide how you will organize the material. A simple table can connect each research question to its method, participant group, and intended output. This makes gaps visible before the work begins. If a question has no plausible source of evidence, narrow it or remove it.
Plan how findings will become a decision
Research is most useful when the team agrees in advance how it will review the results. Decide who needs to see the findings and what form will help them act: a brief summary of themes, a comparison of options, or a walkthrough of observed difficulties. Keep observations distinct from interpretations. “Several participants looked for pricing before signing up” describes what happened; “pricing is the only barrier” goes further than that observation supports.
Return to the original decision when presenting conclusions. State what the evidence suggests, what remains uncertain, and which next step is reasonable. That step might be a change to test, a question requiring more research, or a choice to leave something as it is. A good plan does not promise a perfect answer. It makes the path from business question to evidence to action clear enough for others to evaluate.
