Marketing research explained: the six steps, the methods, and the brief that produces a decision

Marketing research is the organised work of collecting and interpreting evidence about buyers, competitors and the market so that a specific decision can be made with less guessing. The word that makes it useful is specific. Research that begins with a question like "what do our customers think of us" produces a report. Research that begins with "should we raise the price of the standard plan, and what would we lose if we did" produces a decision. The process has six steps: define the decision, choose a method that can answer it, select who to ask, collect the evidence, analyse it, and write down what you will now do. Marketing research covers the whole marketing problem, including pricing, message, product and channel. Market research, strictly, is the narrower study of a market's size, structure and participants, and most people use the two terms interchangeably.

This page explains the process and the methods. For hiring a firm to run it in business to business markets, see what a research agency delivers.

The six steps

• One. Define the decision, not the topic. Write the sentence "we will decide X by date Y, and this research will tell us Z". Then write what result would make you choose each option. Doing this before you collect anything is what separates research from a survey habit.
• Two. Choose the method that can answer it. Questions about why people behave as they do need conversation. Questions about how many need counting. Picking the wrong one is the most expensive error in the whole process.
• Three. Decide who to ask. Define the population, then get a sample that resembles it. Interviewing only your happy customers answers a different question from the one you asked.
• Four. Collect. Run the interviews, the survey, the test or the desk work, and keep the raw material, because you will want to re-read it with a different question later.
• Five. Analyse. Look for the pattern that holds across sources and the one finding that contradicts what the company believes. The second is usually the valuable one.
• Six. Decide and record. One page: the question, what you found, the decision, and the date you will check whether it was right.

Primary and secondary, qualitative and quantitative

Type	What it is	Best for	Watch for
Secondary, desk research	Evidence that already exists: published statistics, industry reports, competitor material, review sites, search demand data	Sizing a market, understanding structure, checking an assumption cheaply. Always do this first	Dates. Use the publication date and the collection period, not the year you read it
Primary, qualitative	Interviews, group discussions, observation, open ended questions	Why people choose, the words they use, what blocks a purchase	Small numbers describe mechanisms, so do not report them as percentages
Primary, quantitative	Surveys, conjoint and price testing, A and B tests, sales data analysis	How many, how much, which of these, at what price	Question wording, which moves answers more than most people expect

The strongest projects run in that order: desk research to frame the question, a dozen conversations to learn the language and the real decision criteria, then a quantitative instrument written in the buyers' own words to size what the conversations found.

The methods, and what each one answers well

Method	Answers
Depth interviews with recent buyers and recent non buyers	Why they chose, who else they considered, what nearly stopped them
Win and loss review of your own pipeline	Which objections recur and which competitor keeps appearing
Survey of a defined population	How widespread a need, a preference or a complaint is
Price sensitivity testing	Where a price change gains or loses volume, and which features carry the price
Concept and message testing	Which of several descriptions a buyer understands and prefers, before you build or launch
Search and query data	What people type when nobody is watching, and in what volume. See keyword research in five steps
Review and forum analysis	Unprompted language about your category, your rivals and their weaknesses
Competitor evidence	Positioning, pricing, hiring and channel moves, covered in public evidence sources
Observation and usability sessions	What people actually do, as opposed to what they say they do

Writing questions that produce usable answers

• Ask about behaviour, not intentions. "What did you do the last time this happened" beats "would you buy this". Past behaviour is evidence; stated intention is a wish.
• One idea per question. Anything containing "and" is two questions with one answer.
• Remove the lead. "How useful would this feature be" assumes usefulness. "What would you use instead of this" does not.
• Use their words. Take the vocabulary from the interviews. A survey written in company language measures whether people can decode company language.
• Put the demographics last. Start with the interesting question while attention is high.
• Pilot with five people. Every confusing question shows up in the first five, and fixing them there costs nothing.

The five mistakes that waste a project

• No decision attached. If nobody can say what action a result would change, the project will end in a slide deck.
• Asking only customers. Your customers are the people you already convinced. The people who chose somebody else hold the answer you are paying for.
• Sample too small to split. If you intend to compare three segments, each needs enough respondents to compare, so decide the cuts before you field the survey.
• Reporting qualitative findings as numbers. Twelve interviews tell you how the decision works, not what proportion of the market thinks something.
• Research as delay. Where the cost of the decision is lower than the cost of the study, run a small live test instead and let the market answer.

A worked one page brief

• Decision. Whether to launch a lower priced entry tier next quarter.
• What we need to know. Whether the buyers who do not convert today are blocked by price or by something else, and what they would give up for a lower price.
• Method. Desk review of competitor tiers, twelve interviews with lost opportunities from the last six months, then a survey of the defined buyer population with a trade off exercise.
• Who. Decision makers in the target segment, split by company size, with enough in each split to compare.
• Decision rule, written in advance. We launch if price is named unprompted as the main blocker by a clear majority of lost opportunities and the trade off exercise shows a feature set we can remove without harming the main tier.
• Date. Findings in six weeks, decision in seven.

Research feeds directly into who you choose to serve, which is covered in the worked segmentation example, and into the buyer description in building a marketing persona.

Asked and answered: marketing research explained

What is marketing research in simple terms?

Gathering evidence about buyers, competitors and the market, and interpreting it, so that a particular marketing decision is made on facts rather than on opinion. It starts with a decision and ends with an action.

What is the difference between marketing research and market research?

Market research studies a market: its size, structure, segments and participants. Marketing research covers any research that informs a marketing decision, including pricing, message, product, channel and advertising. In everyday use the two terms are swapped freely.

What are the steps of the marketing research process?

Define the decision, choose the method, select the sample, collect the evidence, analyse it, then decide and record. The first and last steps are the ones most often skipped, and skipping either turns the project into a report nobody uses.

What is the difference between primary and secondary research?

Secondary research uses evidence that already exists, such as published statistics, competitor material, review sites and search data. Primary research creates new evidence through interviews, surveys, observation or tests. Start with secondary, because it is faster and it tells you what is worth asking.

How many people do we need to interview?

For qualitative work, keep going until new interviews stop producing new information, which usually happens within about a dozen conversations in one segment. For quantitative work, the number is set by the comparisons you intend to make, so list the cuts first and size each one to support a comparison.

Can a small business do this without an agency?

Yes, and the highest value version is free: ten recorded conversations with recent buyers and recent non buyers, read back for the words they use and the objections they repeat. That alone rewrites most websites and most sales scripts.

Add a source to the research brief: the assistant's answer

Surveys, interviews and desk research all describe what people think. An assistant's reply describes something else, which is what a buyer is told when they ask the question your research is about.

Treat it as a data point with a date. Record the question, the assistant, the reply and the brands named. AI Knows Us captures that for several assistants at once, so the brief can include it next to the other methods.