Saturation: When to Stop Interviewing

WHY IT EXISTS: Qualitative research generates rich, non-numerical data, but without a stopping rule, studies can balloon indefinitely. Researchers needed a principled way to know when they had mapped the conceptual landscape of a phenomenon rather than merely collecting more anecdotes. Saturation emerged as that boundary. It replaces arbitrary sample sizes with an empirical signal from the data itself, protecting both timeline and budget while preserving rigor.

THE MENTAL MODEL: Think of saturation like wringing out a wet towel. The first few twists release a lot of water. As you keep going, the drips become smaller and less frequent. Eventually, no matter how hard you twist, nothing new comes out. Saturation is that dry point in data collection. It does not mean you have heard identical stories; it means you have seen the full range of patterns and no new conceptual categories are appearing.

HOW IT WORKS: Saturation is reached through iterative cycles of collection and analysis, not by counting heads. After each interview or observation, the researcher codes the data and looks for emergent themes. When consecutive participants add no new codes, dimensions, or relationships to the framework, the study has likely saturated. In practice, this means overlapping responsibilities between recruiting and analysis. A common benchmark in homogeneous populations is that saturation often appears between twelve and twenty interviews, but the exact number is always contingent on the complexity of the topic and the diversity of the sample.

WHEN TO USE IT: Use saturation in exploratory or interpretive research where the goal is to understand experiences, mental models, or workflows. It fits semi-structured interviews, contextual inquiry, ethnography, and grounded theory studies. It is especially valuable in UX research when mapping user problem spaces before committing to a solution, because it tells you when you have seen enough of the territory to design with confidence.

WHEN NOT TO USE IT: Do not use saturation as a stopping rule for quantitative methods, surveys, or unmoderated usability tests that rely on statistical inference. It is also inappropriate when the research question requires demographic representation rather than thematic depth. A dangerous misuse is treating a small handful of similar participants as sufficient; five identical interviews do not equal saturation, they equal replication bias.

ONE CANONICAL EXAMPLE: Imagine a UX researcher investigating how nurses interact with a medication administration app during night shifts. The researcher interviews nurses across three hospitals, varying tenure and ward type. After the twelfth interview, themes such as workaround scanning, alert fatigue, and handoff workarounds keep repeating with no new subtypes or contextual variations emerging. The researcher stops recruiting, confident the pattern space is adequately mapped, and moves into synthesis rather than scheduling five more identical conversations.