Market research fails in the planning, not in the field. A tight plan starts from the decision the research feeds and works back to the methods that can actually move it. This is the step-by-step method we use.
Step 1: Write the decision the research feeds
Research is a means, not an end. State the decision it supports: "Do we enter this market?" "Is this acquisition thesis valid?" Every hypothesis and method traces back to that decision. Research that answers a question nobody asked is expensive decoration.
Write the decision as a sentence with an action and a threshold: "Enter the UK market only if we can reach 5% share in three years" beats "Understand the UK market". The threshold matters because it gives you a pass/fail test to aim at. If the research cannot move that threshold one way or the other, it is not earning its budget. Put the decision sentence on the front of the plan and keep it visible through fieldwork, because scope creep always starts with a researcher finding something interesting that is not decision-relevant.
Step 2: Turn the decision into hypotheses
Break the decision into testable hypotheses: "The customer pain is real and urgent", "The total market is large enough", "The competitor response will be slow". Each hypothesis gets its own evidence plan. This is what turns research from exploration into validation.
Write each hypothesis so it can be confirmed or rejected, with the evidence it would take to do so. "The pain is urgent" becomes "at least 30% of respondents report the problem at least monthly and rank it in their top three issues". "The market is large enough" becomes "the served market exceeds $40 million across our three target segments". For each hypothesis, write the evidence that would support it and the evidence that would kill it. A plan where every hypothesis survives no matter what is found is a plan that will produce a report that agrees with the client. That is the failure mode to design against.
Step 3: Choose the methods that test each hypothesis
Match methods to hypotheses. Interviews are for understanding why; surveys are for measuring how many; secondary data is for sizing. Don't run a survey to discover a reason, and don't run interviews to prove a number. Each method earns its place by testing a specific hypothesis.
Here is how the split usually falls. Secondary data sizes the market and the segments. Expert and customer interviews test the hypotheses about motivation, pain, and the buying process. A survey then measures how representative those motivations are across the population, because a small set of passionate voices can make a niche look like a trend. Desirability and willingness to pay are best tested with a small pilot or a concept test, not with an attitudinal survey question, because what people say they will pay and what they pay are different numbers. Each method has a cost per insight and a failure mode; write both down before you start.
Step 4: Plan the participants explicitly
For interviews, name the roles and the target count per segment. For surveys, define the sample and what would make it representative. "Talk to some customers" is not a plan; "20 interviews across the top-10 customers plus 5 lost-deal interviews" is.
Worked example: you need to understand why customers leave. Plan 15 interviews with current customers who have renewed, 10 with customers who churned in the last 6 months, and 5 with sales reps who handled the lost deals. That is 30 interviews across three perspectives, and each perspective earns its place because each one tests a different hypothesis about churn. For the survey, think in terms of statistical confidence. At a 95% confidence level with a 5% margin of error you need roughly 385 responses for a large population; for a B2B population of 1,200 buyers, that falls to about 290. If the budget only supports 150 responses, say so up front and accept a wider error band, rather than pretending 150 is representative. Decide response rates realistically too: cold email to buyers returns maybe 2% to 5%, so plan to contact five to ten times more people than the response count you need.
Step 5: Define the secondary sources up front
List the data sources before you start: industry reports, government and registry data, competitor materials, financial databases. The secondary layer does the sizing and framing; the primary layer does the insight. Knowing which is which stops you from burning interviews on facts you could have read.
Name the source next to each data point you plan to collect: market size from a trade association yearbook, company counts from a registry, pricing from a competitor price scan, financial benchmarks from a databases subscription. Assign a responsibility for each source and a status. When the plan is written this way, the fieldwork brief is narrower and cheaper, because the researcher is not spending interviews rediscovering what a yearbook already says. It also makes the sourcing audit trivial later, which matters when the client asks where a number came from.
Step 6: Decide how you will analyse before you collect
Pre-commit to the analysis approach: how findings will be coded, triangulated and turned into conclusions. If you decide the analysis method after the data arrives, you will be tempted to fit the analysis to what you found.
For interviews, agree the coding frame in advance: the themes you expect, the ones you will listen for, and the rule for when a quote becomes evidence. For the survey, pre-specify the segments you will cut the data by and the cross-tabs you will run. Decide in advance how you will handle disagreement between sources: if the secondary data says the market is growing and the interviews say it is flat, which one wins and why? A triangulation rule that is written before the data arrives is a neutral referee. The same rule prevents the report from quietly selecting whichever source supports the preferred conclusion.
Step 7: Plan the timeline and the budget
A research plan without a calendar is a wish list. Sketch the weeks: secondary data in the first week, interviews in weeks two and three, survey fielding in week four, analysis in week five. For interviews, book a full week of calendar for scheduling; executives confirm and cancel late, so build in 30% slack. For a survey, the fielding window itself is rarely the constraint; the analysis is. The budget splits roughly as a third sourcing, a third fieldwork, and a third analysis, and if any of the three is squeezed to nearly nothing, the report will show it.
Step 8: Define the deliverables
Name the outputs: a findings memo, a validated market size, a competitor map. Deliverables are how research becomes a decision, so they need owners and dates.
For each deliverable, say what it must contain to be usable. The findings memo needs the hypothesis verdicts, not a transcript dump. The market size needs the source next to every number. The competitor map needs the dimensions and the evidence behind each placement. Assign an owner and a date to each deliverable, and tie the last one to the decision gate from Step 1, so the research ends with a decision, not a handover.
Speed it up
Plan it with the Market Research planner, which structures the objective, hypotheses, methods and sources, and exports to a deck or Word report.