Exploratory marketing research: what to do before a survey

Exploratory research is the first stage of a marketing research project: you talk to knowledgeable people and read what already exists until a vague worry becomes a clear problem with hypotheses you can test. Exploratory marketing research comes before descriptive research (surveys that count things) and causal research (experiments that prove cause and effect), and its job is to make sure the survey asks the right questions.

Skip it and you get a beautifully designed survey about the wrong thing, with a confidence interval to match.

Marketing research has three designs, and exploratory usually comes first

Marketing research textbooks split research design into three types. A Penn State marketing research course puts the purposes this way: exploratory research obtains background information that leads to hypotheses, descriptive research measures the state of a variable, and causal research tests hypotheses about the relationship between two variables.

Design	The question it answers	Typical methods
Exploratory	What is actually going on, and what should we test?	Expert interviews, in-depth interviews, focus groups, secondary data, past cases
Descriptive	How many, how often, who, where?	Surveys, sales and customer records, panels
Causal	If we change X, does Y move?	Experiments, such as a price change in one week and not the next

The same course says exploratory research typically precedes descriptive research, which precedes causal research, though the order can vary and a causal result can send you back to explore again. Descriptive research measures how often something occurs, and causal research leans on experiments, which the course calls the cleanest way to measure causal relationships.

If you have read our post on generative research, the idea will feel familiar. Generative research is the product and UX framing: it asks what to build. Exploratory research is the marketing research framing: it is a formal first stage of a study, and its output is a problem statement and a set of hypotheses that the descriptive stage will measure. If you are deciding what product to build, read that post. If you are about to commission a survey and are not sure what it should ask, this one is for you.

Exploratory research turns a vague worry into hypotheses

The Penn State course says the purpose of exploratory research is to familiarize the researcher with potential marketing problems, including collecting background, defining terms and sharpening the problem into hypotheses. It also notes the work is informal and relatively unstructured, with often no fixed number of respondents.

That informality is the point. At the start you do not know which questions matter, so a fixed questionnaire would lock in your guesses. A conversation lets the other person tell you about the thing you did not know to ask.

A useful hypothesis is specific enough to be wrong. "Customers are unhappy" is a mood. "Restaurants that lease their dish machine get detergent bundled with the lease" is a hypothesis, because a survey can count how many do.

Here is a made-up example: a detergent subscription nobody renews

Dana runs marketing at a company that sells a monthly detergent subscription for commercial dishwashers to independent restaurants. Sign-ups look fine, but most restaurants cancel after the first box. Her boss wants a survey of 300 restaurants by the end of the month.

Dana does not yet know what to ask. So she spends two weeks exploring first.

She starts with secondary data: the company's own cancellation notes and trade association material on how restaurants buy supplies. Then she books two expert calls with former sales reps at a foodservice distributor, and nine interviews with restaurant owners and kitchen managers, run in rounds of three.

Round one (interviews 1 to 3) surfaces five distinct reasons for cancelling.
Round two (interviews 4 to 6) adds two more, for seven in total.
Round three (interviews 7 to 9) adds none. Every story fits one of the seven.

Two reasons come up most. Six of the nine people say their dish machine is leased and the leasing company already supplies chemicals. Three say they order everything through one distributor and a separate box is one more invoice. Those two become her hypotheses, and the seven reasons become the answer options on her survey question about why restaurants stop.

Notice what the counts are for. "Six of nine" tells Dana which hypothesis to test first. It does not tell her that two thirds of restaurants lease their machines. That number is what the descriptive survey is for.

Four methods do most of the exploratory work

The Penn State course lists focus groups, case analysis, experience surveys and secondary data analysis as exploratory methods. In practice, use them in this order.

1. Read secondary data first

Secondary data is information gathered for some other purpose: government statistics, trade reports, your own sales and support records. It is cheap and fast, and it keeps you from spending an interview on a question a report already answers. Our guide to secondary market research covers where to look and how to tell when it has run out.

2. Talk to experts who know the whole market

What textbooks call an experience survey is a series of interviews with people knowledgeable about the subject. Today most people call these expert interviews. A former distributor rep has seen hundreds of restaurants buy supplies, so two calls can map the territory faster than ten customer conversations. Use experts for the map and customers for their own reasons. Our expert interview guide covers how to run one.

3. Interview customers one at a time

In-depth interviews are where hypotheses usually come from, because one person can walk you through what they actually did and why. Ask about the last time it happened, not what they generally do.

4. Use small groups and past cases sparingly

A focus group is an interview with a small number of people at once. It is quicker than separate interviews, and it makes any one person's reasons harder to hear, so use it to see which language people share. Case analysis, a review of an earlier situation that resembles yours, helps when one exists. The course warns that you need to acknowledge the differences between cases. (Your competitor's failed launch in 2019 had a different price and possibly a different reason.)

Stop exploring when you can write the survey

Exploration has no natural end, which is why some teams never leave it. Switch to descriptive research when all four of these are true.

You can state the problem in one sentence that names a decision. "Find out why restaurants cancel so we can decide whether to sell through leasing companies" is ready. "Learn more about restaurants" is not.
Each hypothesis is specific enough to be wrong. If no survey answer could disprove it, keep exploring.
A round of interviews adds nothing new. Researchers call this saturation. A systematic review of 23 studies on saturation found that the ones testing real interview data reached it within 9 to 17 interviews, or 4 to 8 focus groups, mostly with fairly similar participants and narrow objectives. Broader groups need more.
The questions left are about how many. If what you still want to know is "how common is this?", interviews are the wrong tool and a survey is the right one.

The fourth signal is the most useful. When your notes are full of "I wonder how many restaurants do that", the exploratory stage has done its job.

Short version: stop when your interview answers have become your survey's answer options.

Exploratory findings are directions, not numbers

Exploratory samples are small and chosen for knowledge, not to represent the market. Report them as counts of people ("six of nine") and never as percentages of a market. Our post on qualitative market research covers how to write those findings up.

The other common mistake runs the other way: skipping straight to a survey because it feels more rigorous. A survey can only measure the options you thought to include. If Dana had written her survey on day one, "our machine is leased and chemicals come with it" would not have been on the list, and 300 restaurants would have picked "too expensive" because it was the closest option.

Your next step

Write your research problem in one sentence, list what you would need to believe for each possible decision, and book two expert calls and the first round of three customer interviews. Instant Expert can find people matching a description such as "former sales reps at foodservice distributors who sold to independent restaurants" and send them paid invitations for a short call, and you pay only if someone books. For help designing the study itself, the directory pages for consumer insights experts and marketing consultants are a good start.