How many interviews are enough for a qualitative study?

The honest answer is that there is no number, and the reason people want one is that a defensible sample size has to be argued rather than looked up. The commonly cited figures come from empirical studies of when new codes stopped appearing in particular datasets, and they transfer only to studies resembling those datasets.

It helps to know where the familiar figures came from. Guest, Bunce and Johnson analysed 60 interviews and found thematic saturation by interview 12, on a homogeneous sample with focused aims, and the authors warned that broader questions or more varied samples would need more. Green and Thorogood reported that little new information appears after about 20 people when the question is specific and participants belong to one analytically relevant category (Vasileiou et al., 2018, pp. 2-3). Both numbers describe a narrow design, so quoting them without matching that design is the mistake reviewers catch.

What drives the number is variation. If your participants share a role, a setting, and an experience, a small number covers the range quickly. If you are sampling across three professions and two countries, you need enough in each group to see within-group patterns, and the total climbs fast.

The published counts move with that variation. Hagaman and Wutich needed 20 to 40 interviews to saturate meta-themes across multiple sites and cultures, and Francis and colleagues reached saturation on all their predetermined constructs at interview 17, having set an initial analysis sample of 10 followed by a stopping rule of three further interviews yielding nothing new (Vasileiou et al., 2018, p. 3). Discipline conventions sit in the same range, with 20 to 30 interviews recommended for grounded theory and 15 to 30 for single-case work, against a practical ceiling of about 50 to keep the analysis manageable (Vasileiou et al., 2018, p. 2). Those figures are not interchangeable, and picking the one that suits your recruitment is not a justification.

Design depth pulls the other way. Studies that analyze a small number of interviews in great detail, such as interpretative phenomenological analysis, work with six to ten by design, and adding participants would make the analysis shallower rather than stronger. Naming the tradition you are working in does much of the justification for you.

Practical constraints are legitimate and should be stated rather than dressed up. "Recruitment closed at fourteen because the funded period ended, which limits the range of settings represented" is a defensible sentence. An invented methodological reason for a practical limit is not.

How do I know when I have reached saturation?

When successive interviews stop producing new codes rather than new words. Track it: record how many new codes each interview generates, and watch the curve flatten. Saturation claimed without that record is an assertion, and it is increasingly challenged in review.

Saturation is also relative to your question. You can be saturated on your central theme and thin on a secondary one, so check it per theme rather than for the study as a whole.

Saturation also comes in two forms, and they stop at different points. Hennink, Kaiser and Marconi separated code saturation, the point where no new issues appear, from meaning saturation, the point where further interviews stop adding nuance and dimension to what you already have. They hit code saturation at nine interviews and meaning saturation only at 16 to 24 (Vasileiou et al., 2018, p. 3). If your analysis has to explain how or why something works, nine interviews will give you the topic list and not the account, so state which of the two your argument needs.

The concept has real critics, who argue that meaning is never exhausted and that saturation claims are often retrospective justifications for a sample size chosen for other reasons. Knowing that literature exists is useful, because an examiner may raise it, and the answer is to report what you actually observed rather than to invoke the term.

How do I justify the number to reviewers?

Give the reasoning, not the norm. State the sample's homogeneity, the breadth of the question, the interview length, the depth of analysis, and the point at which new codes stopped appearing. "Twenty was standard in this field" is the weakest available justification.

Report the recruitment story too: how many were approached, how many declined, and whether the people who declined differed systematically from those who took part. That last point affects how your findings should be read and is rarely reported.

There is a reason examiners have stopped taking the word at face value. Across the qualitative health studies reviewed, saturation accounted for 55% of all sample size justifications, and not one of those claims was substantiated by a procedure reported in the study (Vasileiou et al., 2018, p. 15). The same review argues that adequacy is not only about how much data you collected, but about variety, disconfirming evidence, and discrepant cases, and it asks you to weigh your epistemological position, aims, phenomenon, sampling strategy, data richness, and your own interviewing experience (Vasileiou et al., 2018, pp. 15-16). A short paragraph doing that work outperforms the word saturation on its own.

If you have a target from a funder or an ethics application, say what it was and whether you met it. Consistency between the protocol and what happened is itself a quality signal.