The role of marketing research in decision making: match each decision to the evidence it needs

The role of marketing research in decision making is to replace a guess with evidence before money is committed. Research does not make the decision. It narrows the choices, shows what customers do and say, and tells a team which option has the most support. A study that is not tied to a decision that someone must take is a report that sits in a folder.

This page is about that link between a decision and its evidence. The process of running a study, the six steps and the survey methods, sits on a separate page, Marketing research explained. Here the starting point is the decision, and the work is choosing the cheapest evidence that settles it.

What research does for a decision

Every marketing decision carries uncertainty. Will customers pay this price? Will this message make sense to them? Is this segment large enough to serve? Research does four jobs against that uncertainty.

• It describes. Who buys, how often, through which channel, and what they compare you with.
• It explains. Why a customer chose a rival, why a launch stalled, why returns are high.
• It compares. Option A against option B, tested with the same people under the same conditions.
• It tracks. The same questions asked at fixed intervals so a change is visible when it happens.

A decision maker who knows which of the four jobs they need can brief a study in two lines. A decision maker who asks for "research on our customers" gets a long deck and no answer.

Six decisions and the evidence each one needs

Decision	Question the evidence must answer	Fitting evidence
Which segment to target	Who has the problem we solve, and how many of them are there?	Existing customer data, a segmentation survey, sales call notes
What to build or add	Which need is unmet and how strongly is it felt?	Depth interviews, support tickets, a concept test
What to charge	Which price and feature mix do buyers trade off?	Conjoint or a price sensitivity survey, plus real sales history
What to say	Which message is understood and remembered?	Message tests, search query language, customer interviews
Where to be found	Where does the buyer look first?	Channel survey, search data, win and loss interviews
Whether it worked	Did the change move the measure we set?	Tracking study, holdout test, before and after comparison

Read the table from the left. The decision comes first and the method follows. The reverse order, picking a survey because a survey is easy, is how teams collect data that answers a question nobody asked.

The value test before you commission anything

Ask four questions before a study starts. If any answer is vague, fix the brief first.

1. What decision will this inform, and who takes it? Name the person and the date.
2. What would we do differently for each possible result? If the action is the same whatever the data says, the study has no value.
3. What do we already know? Sales data, search data, support tickets and past studies are free and often settle half the question.
4. What does a wrong decision cost? A study should cost a small fraction of the money it protects.

The second question does most of the work. Write down, in advance, "if more than half of the interviewees say X, we will do A, otherwise B". That sentence turns research into a decision tool.

Where research sits in the decision cycle

Before the decision: exploring

Early work is open. Interviews with ten or so customers, a review of what people write in reviews and forums, and a read of the questions typed into search engines show which problems matter. This stage produces hypotheses, not percentages.

At the decision: choosing

Here the work is narrow and quantitative. Two or three options are tested against a sample that looks like the target buyer. The output is a ranking of options and the size of the gap between them, so the team can see whether the difference is large enough to act on.

After the decision: checking

Once the action is live, a short tracking measure shows whether it did what was intended. This closes the loop and gives the next decision a better starting point. Teams that skip it repeat the same expensive exploration every year.

Common ways research fails to inform a decision

• The decision was already made. The study is used to justify a choice, so contrary results are ignored. Agree the rule for acting before the data arrives.
• The sample is the wrong people. Asking existing customers whether to enter a new segment tells you about the old segment.
• The question tests opinion, not behaviour. "Would you buy this?" overstates demand. Where possible, measure what people did: clicks, sign ups, pre orders, search volume.
• The result arrives after the date. A perfect study delivered a week after the launch decision has no role. Fit the method to the calendar.
• Nobody owns the action. Each finding needs a named person and a next step.

Cheap evidence to use before paying for a study

Many decisions can be settled with data a business already holds. Sales records show which products sell together. Support tickets show where customers get stuck. Search Console shows the exact words people type before they find you. Win and loss calls, held with five buyers who chose a rival, often explain more than a survey of five hundred. For competitor context, a repeatable competitor analysis is a low cost starting point, and a perceptual map turns customer ratings into a picture of where a brand sits against its rivals.

When the decision is large enough to need an outside supplier, how to judge a research agency's method covers what to ask for.

Writing the decision memo

Finish every study with a one page memo with four parts: the decision, the evidence, the recommendation and the next measure. If the memo cannot be written in one page, the study answered too many questions or none. Send it to the person who owns the decision, not to the whole company. The memo is the real output of the project, and the deck is its appendix.

What an assistant says when a buyer asks about your category

Customer research now has a new source: the answers a buyer receives when they ask ChatGPT, Gemini or Perplexity to compare options in your category. Those answers shape the shortlist before anyone visits a website, so they belong in the evidence for a positioning or channel decision. Ask the questions your buyers ask, record which brands are named, and note which sites the assistant cites. AI Knows Us (aiknowsus.com) runs the same questions across the major assistants, so a research team can add that view to its tracking set.

Questions about marketing research and decisions

What is the main role of marketing research?

To give decision makers evidence about customers, competitors and markets so that choices on product, price, promotion and place rest on facts rather than opinion.

Is marketing research only for large companies?

No. A small business can run five customer calls, read its own sales and search data, and test two messages on a small paid audience. The method scales down because the logic is the same: name the decision, then gather the cheapest evidence that settles it.

How much should a business spend on research?

Spend in proportion to the cost of getting the decision wrong. A pricing change on the whole range justifies a proper study. A choice between two email subject lines justifies a split test and nothing more.

What is the difference between research and analytics?

Analytics reports what happened on your own channels. Research also reaches people who have not bought from you and explains why. Most decisions need both.

How do I know the research is good enough to act on?

Check that the sample matches the buyer, that the question measured behaviour or a clear choice, and that the gap between options is large enough to matter to the decision rule you set in advance.