“How many people should I interview?” is a familiar question in qualitative research. The most attractive answer is often a number: 12, 15, or 20 participants. Yet no number automatically makes a study adequate. A narrow question studied in a relatively homogeneous group may be addressed with a small sample, while a study involving several groups, settings, or a need for deeper explanation will require more data.
Instead of seeking one threshold for every study, researchers need to demonstrate that their sample fits the question, recruitment strategy, data type, and level of conclusion they intend to make. The number should be an outcome of the design, not a substitute for it.
Ask first what the study needs to know
If the purpose is to identify major issues for questionnaire development, data collection may stop when additional interviews no longer add important codes or themes. If the purpose is to understand how a phenomenon unfolds, why groups interpret it differently, and which conditions change the experience, merely identifying themes is not enough.
Hennink, Kaiser, and Marconi distinguish code saturation from meaning saturation. In their study, the main thematic domains appeared relatively early, but more interviews were needed to understand their depth, nuance, and relationships. The distinction matters: having heard the range of issues is not the same as having understood those issues fully.
Consider how homogeneous the sample is
Participants with the same role, organization, and type of experience usually produce a narrower range of data than a sample spanning industries, seniority levels, or regions. If the study compares new employees with experienced managers, individual with business customers, or users across countries, each group needs enough material for separate analysis and comparison.
The total number of participants therefore does not fully represent design strength. Twenty interviews divided across five groups are different from twenty interviews within one homogeneous group. The sampling plan should state which groups must be represented, why they provide necessary perspectives, and the expected minimum for each group.
Do not use saturation as a final assertion
Many reports simply state that “data saturation was reached” without showing how it was monitored. A credible claim needs a process. Researchers should analyze while collecting data, update the coding framework, record new themes, and define a stopping criterion before collection ends.
A practical method is to maintain a table after each interview: new codes; new dimensions of existing codes; contrary cases; missing groups; and questions to revise. A team may use a criterion such as no important new code in three consecutive interviews, but that criterion fits only a goal of thematic coverage. If the study needs deeper explanation, the team must also examine whether the meaning of each theme is clear and whether contrary cases have been understood.
Interview quality affects sample size
Ten information-rich interviews are not equivalent to ten short conversations based on closed questions with little probing. Interviewer skill, participant relevance, the clarity of the interview guide, and trust within the conversation all affect the amount and quality of data obtained. Adding participants cannot compensate for weak questions.
Before expanding the sample, check three things: whether participants genuinely have relevant experience; whether the questions invite concrete situations rather than general opinions; and whether the interviewer probes causes, sequences, and consequences. A small pilot often improves the guide and prevents thin data during the main study.
Use numerical guidance within its conditions
Hennink and Kaiser’s systematic review found that many empirical studies reached saturation within 9 to 17 interviews or 4 to 8 focus-group discussions. These results, however, occurred mainly when study populations were relatively homogeneous and objectives were narrowly defined. Multi-country studies, studies with several groups, or studies seeking meaning saturation often required larger samples.
These ranges can support an initial budget and schedule, but they should not become a rigid rule. A sound proposal might state: plan 15 to 20 interviews across two groups; analyze concurrently; review code coverage and meaning depth after interview 12; continue recruitment if important themes remain unclear or comparison groups lack sufficient data.
Report sample size as an argument
In the methods section, explain four elements: the sampling logic; planned and actual participant numbers; how data adequacy was monitored; and why collection stopped. If a group was difficult to recruit or a theme did not reach the desired depth, report the limitation rather than using the word “saturation” to conceal the gap.
A qualitative sample does not need to resemble a survey sample, but it still needs transparent and testable justification. The right question is not only “how many people are enough?” It is “enough for which conclusion, about which group, and on what evidence?” Answering those three parts gives the number meaning.
References
Hennink, M., & Kaiser, B. N. (2022), Sample sizes for saturation in qualitative research: A systematic review of empirical tests.
Hennink, M. M., Kaiser, B. N., & Marconi, V. C. (2017), Code Saturation Versus Meaning Saturation: How Many Interviews Are Enough?.
