How Many User Interviews Do You Really Need? A Practical Guide
Stop guessing your interview sample size. Learn the evidence-based framework for choosing the right number of user interviews, from the '5 user' rule to the difference between code and meaning saturation.
The '5 User' Rule: Where It Comes From and Where It Fails
The right number of user interviews is not a single magic number; it is a function of your research goal and audience. The most common number cited, five, comes from Nielsen Norman Group research on usability testing. It is a specific guideline for evaluative research that is often misapplied to other types of qualitative work.
The '5 user' rule originates from a 2000 analysis showing that five participants will reveal about 85% of the usability problems in a design. The core idea is one of diminishing returns: the first user reveals a flood of issues, the second reveals many new ones, but by the sixth user, you are mostly observing problems you have already seen. The recommendation was to run many small, iterative tests of five users, fixing the identified issues between each round.
This rule is effective for its intended purpose: finding flaws in an interface. However, it was never meant for exploratory or generative research, where the goal is to understand needs, behaviors, and motivations. Applying the five-user guideline to discovery research is a common mistake that leads to acting on anecdotal evidence rather than stable, representative themes.
Key Factors That Determine Your Interview Count
The appropriate number of interviews for your study depends on three key factors: your research goal, the diversity of your audience, and the level of project risk. As multiple studies on qualitative methodology confirm, the ideal sample size is guided by the specific aim and information power of the study, not a universal number.
Research Goal
The purpose of your study is the primary driver of sample size. Exploratory research, aimed at discovering new opportunities or understanding complex workflows, requires a larger sample to reach saturation. In contrast, evaluative research, like testing a specific design, can use smaller, iterative samples because the scope of inquiry is much narrower.
Audience Homogeneity
The more diverse your user base, the more interviews you will need. A study on a single, homogenous group (e.g., small business accountants) will require fewer interviews than a study on a product with multiple distinct segments (e.g., teachers, students, and administrators). Each segment needs its own set of interviews to ensure the patterns you identify are not just individual quirks, as participant variability is a key factor in reaching saturation.
Project Risk
High-stakes strategic decisions demand greater rigor and a larger sample size. If the research informs a decision to enter a new market or build a major new product, the cost of being wrong is high. A larger sample provides more confidence in the findings. For low-risk decisions, like tweaking an existing feature, a smaller sample may be sufficient to provide directional insight.
A Framework for Choosing Your Number
For exploratory research, aim for saturation, which academic studies have shown occurs in stages. For evaluative research, the '5 users per round' rule is a good starting point. For studies involving distinct segments, plan for at least 6-8 interviews per segment to find stable patterns.
For Exploratory and Generative Research
The goal is to reach saturation, the point where new interviews stop yielding new insights. A foundational 2006 study by Guest, Bunce, and Johnson published in Field Methods found that for a homogenous group, code saturation—the point where no new high-level topics emerge—occurred within the first 12 interviews. In fact, they saw the core of major themes in as few as six.
More recent research adds a crucial distinction. A 2017 study in Qualitative Health Research by Hennink, Kaiser, and Marconi separated code saturation from meaning saturation. They found that while you might identify all the topics in as few as nine interviews (code saturation), achieving a deep understanding of their nuances and dimensions (meaning saturation) requires between 16 and 24 interviews. For most business research, this provides an evidence-based range: aim for 9-12 interviews to identify key themes, and 16-24 to understand them deeply.
For Evaluative and Usability Research
Here, the '5 users per round' rule is a practical guideline. The key is the phrase “per round.” The intention is not to run five tests and stop, but to conduct iterative cycles of testing, fixing, and re-testing. Five users are enough to find the most obvious problems, which you can then solve before testing the improved design with another small group of participants.
For Segmentation and Persona Development
When your research needs to compare distinct groups, you must plan your sample on a per-segment basis. A common mistake is to conduct 15 interviews across three segments and assume you have enough data. Instead, you need to reach saturation within each group independently. A practical starting point is to plan for 6-8 interviews per defined user segment. For example, a study comparing new users to power users would require 12-16 interviews total (6-8 from each group) to ensure the findings for each segment are reliable.
Working with Real-World Constraints
When time and budget are limited, it is better to narrow the scope of your research question than to cut the number of interviews needed to answer it well. Be explicit with stakeholders about what a smaller sample can and cannot prove, and look for ways to make research more efficient without compromising quality.
Instead of asking a broad question like “How do enterprise customers manage their data?” with only five interviews, narrow the scope to “How do enterprise data analysts in the finance industry complete their first-of-month reporting?” A narrower question can be answered confidently with a smaller, more focused sample.
It is also critical to manage stakeholder expectations. Frame findings from a small sample as directional insights or hypotheses, not definitive conclusions. Explain the risk of over-generalizing from a handful of conversations.
Scaling research to reach saturation with confidence is also becoming more practical. Platforms like Pulse run AI-moderated interviews at a scale that was previously impossible, allowing teams to complete dozens or even hundreds of interviews overnight. This allows you to gather sufficient evidence without the traditional time and cost constraints.
Frequently Asked Questions
What is thematic saturation?
Thematic saturation is the point where new interviews stop yielding new insights. Researchers distinguish between 'code saturation' (when you have heard all the high-level topics, often in 9-12 interviews) and 'meaning saturation' (when you deeply understand those topics, often needing 16-24 interviews).
How do I know if my audience is 'homogenous'?
A homogenous audience shares similar contexts, goals, and behaviors relevant to your research question. For example, a group of surgeons using a specific medical device is homogenous. An audience of 'small business owners' is heterogeneous because it includes people from different industries, company sizes, and technical skill levels.
Does the number of interviews change for B2B vs. B2C research?
Not inherently, but the context often does. B2B research frequently involves more distinct user roles (e.g., buyers, administrators, end-users) that function as separate segments. This segmentation is what increases the total number of interviews needed, as you must study each role to understand the complete picture.
Can I use fewer interviews if I also run a survey?
No, they serve different purposes. A survey is a quantitative method that answers 'what' and 'how many' across a large sample. Interviews are a qualitative method for understanding 'why'. A survey can validate the prevalence of a theme found in interviews, but it cannot replace the deep insight that comes from a conversation.