How Many Interviews Do You Need for Saturation? Why That’s the Wrong Question

Every qualitative researcher has heard the rules of thumb: twelve interviews, fifteen, twenty, “until saturation.” And every honest researcher knows the discomfort of writing “saturation was reached” without being able to say exactly what that meant. The literature confirms the problem: claims of saturation are often made without specifying the criterion used to support the claim, and rule-of-thumb estimates dominate practice. Reviewers have noticed.

The real question

Sample-size guidance answers a logistics question: how much data should I collect? Saturation answers an epistemic question: have I fully used the data I have? These are different questions, and conflating them is why so many methods sections are vulnerable. A study with thirty shallow interviews can be far less saturated than a study with twelve deeply analysed ones. Charmaz made the point years ago: saturation pertains to the comprehensiveness of the theory, not the size of the sample.

Shifting saturation into the analysis

The PRICE model (Naeem et al., 2024) defines saturation as the point at which the researcher has exhausted all possible uses of the gathered data — a cessation in the emergence of new patterns, codes and themes, indicating the data’s utility in addressing the research questions has been maximised. It operationalises this through five processes: Perspectivation (viewing the data through every relevant lens — research questions, theoretical and epistemological frameworks), Recapitulation (summarising findings as emerged codes and themes), Integration (connecting and making sense of them), Crystallisation (structuring them into figures and tables), and Edification (showing how the result improves understanding of the topic).

Three practical tests you can apply today

First, the keyword frequency test: track how many times keywords are repeated by participants using the same word or different words with similar meaning. The velocity of keywords (how often they repeat) and their viscosity (the richness of different words used for the same narrative) tell you whether new data is still adding meaning. Second, the perspective test: are your quotations drawn from a wide range of participants, or are you repeatedly citing one or two fruitful interviews? Cherry-picked quotes from a small number of voices are a saturation red flag. Third, the concept test: when you attempt Crystallisation — putting your connected themes into a figure — does anything in your data fail to find a place, or does any box lack data? Either gap means you are not yet saturated.

What to write in your methods section

Instead of “saturation was reached after N interviews,” report saturation as an analytical achievement: describe how perspectives were exhausted, how codes and themes were summarised and integrated, how the final structure crystallised, and what the framework contributes. That is a claim reviewers can verify — and one they rarely challenge, because the evidence is on the page.