- The cost of a market study is driven mainly by the number of interviews, the data collection method, the questionnaire length and the geographic coverage.
- A face-to-face interview costs several times more per respondent than a telephone one, and many times more than an online one, because of travel, field time and verification.
- Quality control, analysis and reporting are fixed costs relative to the sample; removing them saves little and loses much.
- Cost is reduced properly by narrowing the population, shortening the questionnaire, reducing the reporting subgroups and matching the method to the population, not by cutting the sample below the threshold of usefulness.
- The most expensive research is the cheap study whose results cannot be used.
1. The wrong question and the right one
"How much does a survey cost?" cannot be answered, any more than "how much does a building cost?". A telephone survey of 300 interviews with a company's customers and a national face-to-face survey of 1,500 interviews are both "a survey" and differ in price by a factor of ten or twenty. The right question is: what determines the cost of market research, and which of those factors can be adjusted without compromising the result?
This guide gives no prices, because they depend on the market, the moment and the agency. It gives the structure of the cost instead, so that a proposal can be read, compared and negotiated with understanding. For preparing the brief that produces comparable proposals, see the guide How to write a research brief.
2. What the cost is made of
The cost of any quantitative study breaks down into four blocks, and the proportions between them say a lot about the proposal.
| Cost block | What it contains | How it varies |
|---|---|---|
| Design | Defining the design, sampling, questionnaire, programming, pilot, translation | Relatively fixed; depends on complexity, not on the number of interviews |
| Data collection | Interviewers, travel, calls, respondent incentives, field coordination | Proportional to the number of interviews and the length of each; the largest block in CAPI |
| Quality control | GPS and audio checks, re-contacting, data cleaning | A share of collection; 10-20% in CAPI, less in CATI and CAWI |
| Analysis and reporting | Weighting, tables, tests, advanced analyses, report, presentation | Relatively fixed; depends on the depth of analysis, not on the sample |
In a national CAPI study, data collection usually represents 55-70% of the cost. In an online study, collection may be under 30%, and design and analysis become the main blocks. The practical consequence: in CAPI, cost is reduced through sample and length; in CAWI, cutting the sample saves little, because that is not where the cost is.
3. The six factors that move the price
The number of interviews
The first factor, but not a linear one. The cost per interview falls with volume, because design and analysis are spread over more interviews. The sample size is set by the reporting subgroups and the precision required, and the methodology FAQ explains the calculation. A sample of 1,100 people gives a margin of ±3% in total; if results must be reported for five regions with the same precision, the sample needed is five times larger, and so is the budget.
The data collection method
The second factor and the one with the widest range. As an order of magnitude, a face-to-face interview costs three to five times more than a telephone one, and a telephone interview two to four times more than an online panel interview. The difference comes from human time: the interviewer's travel, call-backs to selected households, the length of the interview, the checks. The method is not chosen by price but by population; but when the population allows several methods, the cost difference is decisive.
Questionnaire length
A 40-minute questionnaire costs almost twice as much as a 20-minute one in CAPI and CATI, because interviewer time is the main cost. Long questionnaires also have higher drop-out rates and need more contacts for the same number of completed interviews. Every question in the questionnaire must have a use in the report; the "nice to have" ones are paid for.
Geographic coverage and the accessibility of the population
A national sample with points in small rural localities costs more per interview than an urban one, because of travel and the low density of respondents. A hard-to-find population, buyers of a niche category, users of a specific service, requires screening: to find 500 eligible respondents, interviewers may contact 5,000 people, and the unsuccessful contacts are paid for.
Quality control
Verifying 15-20% of CAPI interviews through re-contact and audio adds a visible cost to the proposal. It is the block that cheap proposals cut first, because its absence does not show in the report. It shows in the results, but only when they contradict reality.
Depth of analysis
Frequency tables and cross-tabulations cost little; segmentation through cluster analysis, driver analysis, price modelling or perceptual maps require qualified analyst time. The report itself ranges from annotated tables to a strategic document with recommendations, and the presentation and interpretation workshop are separate components.
4. The cost of qualitative research
For focus groups and in-depth interviews, the structure is different: the number of interviews is small, but the cost of each is high. A focus group includes recruiting 9-10 participants with screening, their incentives, the room, the moderator, recording, full transcription of two hours of discussion and analysis. An in-depth interview with an expert or decision-maker includes recruitment, which can take weeks, a senior moderator, transcription and analysis. Cost moves mainly with the number of groups or interviews, the difficulty of recruitment and the language. Saving by reducing the number of groups below three per category is not a saving: a single group allows no conclusion.
5. Where savings are possible without losing validity
- Narrow the population to the relevant one. If the decision concerns buyers of children's products, the general population is not needed; a smaller, cheaper sample of households with children answers better.
- Shorten the questionnaire. Remove the questions with no use in the decision. A 15-minute questionnaire instead of 30 cuts collection cost by up to 40% in CAPI.
- Reduce the reporting subgroups. Reporting by three macro-regions instead of twelve districts reduces the required sample several-fold.
- Choose the method by population, not by habit. A study of existing customers by CATI instead of CAPI costs a fraction and is just as valid.
- Use a mixed design. CAPI for the segments without telephone or online coverage, CATI or CAWI for the rest, with joint weighting. It requires methodological competence, but it cuts cost without losing coverage.
- Do desk research first. What can be learned from statistics is not paid for in the field, and the questionnaire gets shorter.
- Plan waves. A large study every two years and short tracking waves in between cost less than large annual studies and give more information.
6. Where saving destroys the study
- A sample below the threshold of usefulness. 300 respondents for a study that must be reported on six segments means 50 per segment and margins of ±14%; the segment results are noise.
- Online for a population that is not online. The saving is real; the result describes another population.
- No pilot. 20-30 interviews are saved and all the others are put at risk.
- No quality control in the field. Nobody finds out what was invented.
- No weighting and no analysis. Raw tables from an unbalanced sample describe the sample, not the population.
- A questionnaire written in-house to save on design. Leading questions produce reassuring and false results.
7. How to read the price in a proposal
Ask for proposals broken down into the four blocks, not as a single sum. Calculate the cost per valid interview and compare it across proposals with the same method. Check what percentage of interviews is verified and how. Ask what happens to the price if the response rate is lower than estimated: a serious proposal has an answer, usually an additional cost per contact above a threshold. Check what is included in the analysis and what is charged separately: the presentation, the data set, additional tables after delivery.
Two proposals with very different prices for the same brief almost always mean that one of them understood the brief differently or cut a block the other kept. The difference is clarified with a question, before the decision, not with a report, after it.
8. Cost against the value of the decision
The budget of a study is judged against the decision it supports, not on its own. A market study that costs 2% of a launch budget and halves the risk of a positioning mistake is cheap. A cheap survey whose results cannot be presented to the board, the donor or the ministry, because the method does not stand up to questions, is expensive whatever its price, because it is paid for twice: once for itself and once for the proper study that replaces it.
The question to ask before signing is not "is it the lowest price?" but "is it the lowest sum for which I get a result I can use?". For most decisions, the difference between the two questions is the difference between research that cost and research that was worth it.
See also: CAPI and CATI CAWI Online Panel Focus groups and IDI How to write a research brief Face-to-face interviews (CAPI) Methodology FAQ
Do you have a decision to take?
Tell us the problem, not the method. We propose the right research design and explain what it can and cannot deliver.