How many user interviews is enough? How to answer stakeholders who ask whether 8 interviews can support a product decision

You've just finished a round of eight customer interviews. The patterns are clear, and you have a recommendation. Then a stakeholder asks: "Is eight interviews really enough to make this decision?"

It's a fair question, and a common one. Usually it isn't a challenge to your competence. It's a sign that the person asking is used to a different kind of evidence. Product leaders, executives, and engineers often work with dashboards, A/B tests, and survey results, where bigger samples mean more confidence. Eight sounds small next to that.

The good news is that there's a solid, honest answer. Sometimes eight interviews is plenty. Sometimes it isn't. This article explains how to tell the difference, what the research says about sample sizes, and how to respond to stakeholders in a way that builds trust instead of putting people on the defensive.

Why stakeholders ask this question

Most people learn about evidence through a quantitative lens. If a poll surveys 50 people, you wouldn't trust it to predict an election. So when someone hears "we talked to eight customers," they apply the same logic and conclude the sample is too small.

The problem is that interviews and surveys answer different questions:

• Quantitative methods (surveys, analytics, experiments) tell you how many and how much. They estimate how common something is across a population, and they need large samples to do that reliably.
• Qualitative methods (interviews, contextual inquiry, usability tests) tell you what and why. They uncover problems, motivations, mental models, and context. Their job is to discover, not to measure.

When a stakeholder asks whether eight is enough, they're often silently asking a quantitative question ("How many of our customers feel this way?") about a qualitative study. Part of your job is to make that distinction visible without lecturing.

What the research says about interview sample sizes

There's no universal magic number, but there is a substantial body of research you can point to.

Data saturation

The most widely used standard for qualitative sample size is saturation: the point at which new interviews stop producing meaningful new themes. You keep hearing the same problems, described in slightly different ways.

Several studies have tried to measure when saturation happens:

• A frequently cited 2006 study by Guest, Bunce, and Johnson found that most themes in their dataset appeared within the first 12 interviews, and the core themes were already visible after about six.
• A 2022 systematic review by Hennink and Kaiser looked at empirical tests of saturation and found that it was typically reached somewhere between 9 and 17 interviews. Studies with relatively similar participants and focused objectives reached saturation at the lower end.

These numbers aren't rules. They show that for a focused question and a reasonably similar group of participants, a sample in the high single digits to low teens often captures most of what you'll learn from interviews.

The "five users" guideline

In usability testing, you may hear that five users find most usability problems. This comes from work by Jakob Nielsen and Tom Landauer in the early 1990s, which modeled how quickly new problems are discovered as you add test participants.

It's useful, but it's often misquoted. The finding applies to usability testing of a specific interface, assumes problems are fairly easy to encounter, and works best as part of repeated rounds of testing. Later research, including a 2003 study by Laura Faulkner, showed that results from any single group of five can vary quite a bit. It doesn't mean five people can answer any product question.

What this means for you

Eight interviews sits within the range where research shows you're likely to capture the main themes, as long as your audience is focused and your question is narrow. That "as long as" is what you need to explain to stakeholders.

The real answer: enough for what?

The most useful response to "Is eight enough?" is another question: "Enough for what?" Whether a sample is sufficient depends on several factors.

1. How focused is your research question?

"Why do new admins abandon setup on the integrations step?" is a narrow question. Eight interviews with recent admins who dropped off can answer it well.

"What do our customers want from the product next year?" is broad. Eight interviews will surface ideas, but they won't cover all the needs across your customer base.

2. How similar are your participants?

Saturation comes faster when participants share a role, context, and set of goals. If you interviewed eight finance managers at mid-sized companies, that's a solid sample for that group.

If your eight participants included two developers, two designers, two executives, and two support agents, you effectively ran four studies of two people each. That's rarely enough to say much about any one group.

3. What kind of claim are you making?

Interviews are strong evidence for some claims and weak evidence for others:

4. How risky and reversible is the decision?

This is often the most persuasive point for stakeholders. Match the amount of evidence to the cost of being wrong.

• Low risk, easy to reverse: Tweaking onboarding copy, adjusting a workflow, or prioritizing a bug fix. Eight consistent interviews is usually more evidence than these decisions typically get.
• Medium risk: Building a new feature or redesigning a key flow. Eight interviews can justify moving forward, ideally with a prototype test or staged rollout to validate.
• High risk, hard to reverse: Entering a new market, changing pricing, or sunsetting a product. Eight interviews should inform the decision, not settle it. Add quantitative validation and more segments.

5. How strong are the patterns?

Eight out of eight participants describing the same frustration unprompted is strong evidence. Three out of eight mentioning something in passing is a lead to follow up on, not a conclusion.

How to answer the stakeholder

Here's a practical approach you can adapt to the conversation.

Acknowledge the question as reasonable

Start by validating the concern. Something like: "That's a fair question. Eight is a small number, so it's worth being clear about what it can and can't tell us."

This changes the tone of the conversation. You're not defending your work. You're helping someone understand the evidence.

Explain what interviews are designed to do

Keep it short and concrete: "Interviews aren't meant to measure how common something is. They're meant to find out what problems exist and why. Research on qualitative methods shows the main themes in a focused group usually show up within roughly 6 to 12 interviews."

Show your saturation evidence

If you tracked new themes as you went, share that. "By interview five, we stopped hearing new reasons for the drop-off. Interviews six through eight confirmed the same three causes." This is much more convincing than citing studies, because it's about your data.

Separate what you know from what you don't

Be explicit about confidence levels:

• "We're confident this problem exists and we understand why it happens."
• "We don't yet know how many customers it affects. Here's how we could find out."

Stakeholders tend to trust researchers more when they're open about limitations.

Tie the evidence to the decision

Bring it back to risk: "For this decision, which is a change we can test and roll back, I think this is enough to move forward. If we were making a bigger bet, I'd recommend validating with a survey or usage data first."

Offer a next step if needed

If the stakeholder still has doubts, propose a proportionate way to reduce uncertainty instead of arguing. Options include a short survey to measure how widespread the issue is, checking product analytics for behavior that matches what participants described, a few more interviews targeting a segment you didn't cover, or a prototype test before committing engineering time.

When eight interviews really isn't enough

Being honest about limits makes your recommendations more credible. Eight interviews is likely insufficient when:

• You're still hearing new themes in the last two or three interviews.
• Your audience covers several distinct segments and you only have one or two participants in each.
• Participants were hard to recruit and you ended up with whoever was available, not who you needed.
• The decision is expensive or irreversible, and you have no other supporting data.
• Findings are mixed, with no clear pattern across participants.
• The stakeholder needs a number, such as market size or adoption rate, which interviews can't provide.

In these cases, say so before someone else does. "This round gave us strong signals for enterprise admins, but we only spoke with two small business owners. I wouldn't generalize to that group yet."

How to make eight interviews more convincing

The number of interviews is only one part of rigor. The quality of your process matters at least as much, and it's often what stakeholders are really trying to assess.

Recruit carefully

Write a clear screener tied to your research question. Eight of the right people is far more valuable than twenty loosely matched ones. Document who you spoke to and why.

Use a consistent discussion guide

Covering the same core topics with each participant makes patterns easier to spot and compare. Leave room to follow interesting threads, but keep the backbone consistent.

Analyze systematically

Tag transcripts against consistent themes, and track how many participants mention each one. This turns "I heard this a lot" into "six of eight participants described this, unprompted." A customer context platform like Dovetail can help here by keeping transcripts, tags, and highlights in one place, so you can see how often a theme appears and who it came from.

Make the evidence traceable

Every finding should link back to the specific quotes or video clips that support it. When a stakeholder can watch three customers struggle with the same screen, the sample size debate often fades. Seeing real people describe a problem in their own words is persuasive in a way that numbers alone aren't.

Build on past research

A single round of eight interviews rarely stands alone. If earlier studies, support tickets, or sales call notes point in the same direction, bring them in. A searchable research repository makes it much easier to show that today's findings match what you've seen before, which strengthens confidence beyond any single study's sample size.

Quick responses to common pushbacks

"But that's only eight people." "Right. It's not meant to tell us how many customers have this issue. It tells us what the issue is and why it happens. We stopped hearing new themes after interview five."

"How do we know these eight are representative?" "We recruited specifically for recent admins who abandoned setup, which is the group this decision affects. I wouldn't apply these findings to other segments without talking to them too."

"Can we just run a survey instead?" "A survey would be a great next step to measure how widespread this is. The interviews told us what to ask about. Without them, we wouldn't have known which questions mattered."

"What if we're wrong?" "This change is easy to test and roll back. If we ship it behind a flag and watch completion rates, we'll know within a few weeks."

The bottom line

When a stakeholder asks whether eight interviews is enough, they're asking a reasonable question that deserves a clear answer, not a defensive one. The honest answer is that eight interviews is often enough to understand a problem in a focused group, rarely enough to measure how common it is, and never enough by itself for a high-stakes, irreversible bet.

Your job is to make that distinction clear: show your saturation evidence, separate what you know from what you don't, match the evidence to the risk of the decision, and offer a proportionate next step when more certainty is needed. Do that consistently, and the question changes from "Is eight enough?" to "What do we need to learn next?"

No, and it isn't meant to be. Statistical significance applies to quantitative methods that estimate how common something is across a population. Interviews are a qualitative method designed to uncover what is happening and why, such as the problems people face, the reasons behind their behavior, and the language they use. Eight well-recruited interviews with a focused audience can reliably surface the main themes. To learn how widespread those themes are, pair the interviews with a survey, analytics, or another quantitative method.

Data saturation is the point at which additional interviews stop producing meaningful new themes. You keep hearing variations of things you've already heard. Saturation is the most widely accepted way to judge whether a qualitative sample is large enough. It is reached sooner when the audience is similar and the research question is narrow, and later when the audience is diverse or the question is broad. Tracking new themes per interview as you go gives you concrete evidence of saturation to share with stakeholders.

Describe patterns in terms of what you observed, not how common they are in the market. For example, say 'six of eight participants described this workaround' rather than '75% of users do this.' Separate what you're confident about (the existence and nature of a problem) from what you can't yet tell (how many customers it affects). Link each finding to supporting quotes or clips so stakeholders can see the evidence for themselves, and be clear about who you spoke to and who you didn't.