Which type of research solves your problem: a guide for managers and organisations

1. The decision comes before the method

Most research that disappoints was not carried out badly. It answered the wrong question correctly. A manager notices that sales are falling, asks for "a survey" and receives, six weeks later, an 80-page report of tables from which it is impossible to tell what should be done. The problem was not the survey. It was that nobody formulated the decision the survey was meant to support.

Every type of research starts from two distinct formulations. The first is the decision problem: what the manager or the organisation needs to do. The second is the research problem: what information is needed to take that decision, and how it can best be obtained. Falling sales, lost market share or low participation in a social programme are symptoms, not research problems. Research must reach the causes: which customer segments left, where they went, why, and what would bring some of them back.

Before talking to an agency, write down four things:

• The decision about to be taken: launch, reposition, change the price, scale up the programme, withdraw.
• The options on the table, even if there are only two.
• The information that would change the choice between those options. If no information would change the decision, the research is not needed.
• The deadline by which the decision must be taken, because it narrows the possible methods.

With these four elements, choosing the type of research becomes almost mechanical. Without them, every method looks equally suitable, and the choice is made on price or on habit.

2. Three questions that choose the type of research

There are many classifications of market research, but for the person commissioning a study only one distinction matters at the start: what kind of answer is needed. Three questions cover almost every situation.

"How much?" calls for quantitative research

Quantitative research measures. It answers questions such as: how well known is the brand, what percentage of customers would recommend the service, how many residents of a region have heard of a programme, how strong is purchase intention at a given price. The results are figures obtained on representative samples, with a known margin of error, and they can be generalised to the whole population studied. The typical instruments are CAPI and CATI surveys and CAWI online questionnaires.

What quantitative research cannot do is explain. A survey shows that 34% of customers are dissatisfied, but it rarely says, in depth, why. The questionnaire offers predefined answer options, and the real reasons may not be among them.

"Why?" calls for qualitative research

Qualitative research explains. Through focus groups and in-depth interviews, it uncovers the motivations, barriers, associations and the language people use when they talk about a product, an institution or a behaviour. It is the right method when a message generates no reaction, when a good product does not sell, when an information campaign does not change behaviour, or when you do not yet know which questions deserve a place in a questionnaire.

What qualitative research cannot do is give percentages. Eight people in a room do not represent a market, however well they were chosen. The conclusions are deep but not statistically generalisable, and a focus group used as a "vote" produces wrong decisions.

"What is already known?" calls for desk research

Desk research, also called secondary research, uses data that already exist: official statistics, foreign trade data, published reports, prices in online stores, your own sales or CRM data. It is the rational first step in any project, because it is fast and inexpensive and because it establishes what genuinely remains unknown. The size of a market, the trend in imports, the demographic structure of a region or the profile of existing customers can usually be established without a single interview.

What desk research cannot do is tell you what your customers think of you. Existing data describe the market, not the specific perceptions you want to change.

In practice, the three types are combined. A frequent sequence is desk research, then qualitative research to formulate hypotheses and language, then quantitative research to measure them. Another, equally useful sequence is quantitative research followed by qualitative interviews that explain the unexpected results. This is the logic of a mixed-methods design, and a single type of research is rarely sufficient on its own for an important decision.

3. Exploratory, descriptive or causal: the design that shapes the study

Once you know what kind of answer you are looking for, the second choice is the research design. Methodology distinguishes three designs, and each determines the sample, the instrument, the duration and the budget.

Exploratory research is flexible and is used when the problem is not yet clearly defined. It usually precedes a descriptive study and avoids the error of measuring irrelevant things precisely. Descriptive research is structured and measures a state of affairs: once, in a cross-sectional design, or repeatedly, in a longitudinal design, also called tracking, which shows how awareness, satisfaction or trust evolve over time. Causal research is the only design that demonstrates an effect: whether a price cut increases sales, whether an information campaign changes behaviour, whether a social programme produces the result it was funded for. It requires comparable groups and control over the variables, so it costs more, but it is the only design that justifies the statement "the measure worked".

4. From problem to type of research: eight frequent situations

The table below links real situations in companies and organisations to the type of research and the suitable method. It is not a recipe, but the starting point for the conversation with the agency.

The difference between companies and organisations lies not in the methods but in the object. Companies measure markets, customers and competitors. International organisations, ministries and NGOs measure populations, perceptions and behaviour change. Both need probability samples when they want to generalise, and both make the same mistake: they ask for figures before understanding the phenomenon, or they ask for explanations from an instrument built to count.

5. The data collection method: the respondents decide

Once the type of research is chosen, the data collection method remains. Here the decisive criterion is not the budget but the target population: who has to answer, and how they can be reached correctly.

• Face-to-face interviewing (CAPI) suits nationally representative samples, long questionnaires of over 30 minutes, sensitive topics and populations with limited access to telephone or internet, particularly in rural areas and among older people. It is the standard data collection method for social surveys and KAP studies.
• Telephone interviewing (CATI) is efficient for questionnaires of 10-20 minutes, for repeated monitoring and for satisfaction studies among existing customers, where a contact database is available.
• Online surveys (CAWI) are fast and economical for concept tests, campaign evaluations and studies among populations with high internet access. For the general population, the representativeness of an online panel must be checked and corrected through weighting, and some groups remain under-represented regardless.
• The focus group brings together 5-8 homogeneous participants and is run in 3-4 groups for each category studied. It is the method for testing concepts, messages and packaging and for understanding the consumer's language.
• The in-depth interview is the method for experts, decision-makers, B2B customers and intimate topics, where the presence of other people would inhibit the answers.

The practical rule is simple: the population decides the method, and the method decides the cost. The cheapest data collection method applied to the wrong population produces a precise answer to a question you did not ask. Sample size, margin of error and the types of sampling are explained on the methodology FAQ page.

6. Frequent mistakes when choosing the type of research

• The method before the question. "We want a focus group" or "we want an online survey" before knowing which decision has to be supported.
• A survey for an undefined problem. The questionnaire measures precisely what was not important.
• Online for the whole population. An online panel describes internet users well and everyone else badly.
• Subgroups that are too small. A sample of 1,000 people has a margin of ±3% in total, but a subgroup of 80 respondents has a margin of over ±10%, and conclusions about it are almost worthless.
• Customers as forecasters. The question "would you buy this product?" systematically overstates demand. Intention is measured with calibrated scales and interpreted with caution.
• The focus group as a vote. Eight people who prefer option B do not mean the market prefers option B.
• Ignoring existing data. Fieldwork is paid for information that already existed in official statistics or in your own CRM.
• The questionnaire written by the marketing department. Leading and double-barrelled questions produce the desired results, not the real ones.

7. What to ask the agency before signing

A serious research proposal answers the list below without being asked. If a point is missing, request it in writing.

• The research problem restated by the agency in its own words, so you can check that it has been understood.
• The proposed design and the reason it is exploratory, descriptive or causal.
• The target population, precisely defined, and the sampling frame from which respondents will be selected.
• The sample size and the margin of error, in total and for every subgroup you will report separately.
• The data collection method and its justification in relation to the population.
• Quality control in the field: GPS checks, audio recordings, re-contacting respondents, the validation rate.
• The weighting procedure and the source of the reference data.
• The deliverables, their format and the timeline, including the delivery date of the cleaned data set.
• Who analyses the data and who writes the interpretation, not only who collects it.

8. What you receive and how to use the result

A properly completed study delivers a report with a description of the methodology, weighted tables, charts, interpretation of the results and recommendations, accompanied by the data set and a presentation for decision-makers. The part that matters, however, is the return to the starting point: the decision formulated at the beginning. A good report states explicitly which option the data support, with what degree of confidence, and what remains unknown.

If you formulated the decision, the options and the information needed before commissioning the study, reading the report takes an hour and ends with a choice. If not, the report becomes one more document that "confirms it is complicated". Choosing the right type of research is not a matter of method but of clarity about the decision, and that clarity is the responsibility of the person who commissions the study, before it is the responsibility of the one who carries it out.

See also: Quantitative research Qualitative research Data analysis and desk research CAPI and CATI Focus groups and IDI Methodology FAQ

The Perspicuus research team · 26 October 2026