Short answer
There is no single number. You have enough interviews when you have heard from each group whose experience may differ, and new conversations mostly repeat what earlier ones said. For a focused question inside one organisation, that can be ten people. A question spanning sites, roles and shifts needs more.
Whatever the number, report it honestly:
- give counts rather than percentages ("7 of 10 people describe it, 2 disagree, 1 did not know");
- name the groups you did not reach;
- do not present a small study as if it represented everyone.
The question decides the number
| Question shape | What drives the number |
|---|---|
| One process, one site, one role | How many people actually do the work |
| One process, several sites | At least a few people from each site |
| Experiences that differ by shift, contract or tenure | Each group large enough to compare |
| A rare event | Everyone who experienced it, if you can reach them |
A question about how month-end close works in a finance team of eight may need everyone in the team. A question about why new hires struggle across five warehouses needs people from each warehouse. It probably needs both the new hires and the people who train them.
Saturation, used carefully
Qualitative researchers often stop when new interviews stop adding new themes. It is a sensible rule of thumb, with two traps:
- Saturation within one group says nothing about a group you have not asked. Ten day-shift operators can agree with each other and tell you nothing about nights.
- The last few interviews may add the most important point. The people who are hardest to reach often have a different experience.
Keep a simple log as answers arrive: which themes each conversation raised, and whether any were new. Once three or four consecutive conversations in a group add nothing new, that group is probably covered for this question.
Report counts, not percentages
With ten people, "70%" suggests a precision the study does not have. "7 of 10" is honest and easy to check. The AQuA Book is the UK government's guidance on producing quality analysis. It treats uncertainty as something to communicate to the people who use the analysis. In a small interview study, that means reporting:
- the count, out of the number asked;
- how many disagreed, and what they said;
- how many did not know, or were not asked about that topic;
- which groups are missing or thinly covered.
| Weak reporting | Stronger reporting |
|---|---|
| “Most staff feel the rota is unfair.” | “7 of 10 described rotas changed at short notice; 2 said changes were rare; 1 was not asked.” |
| “Communication is a key theme.” | “6 of 12 described not receiving the revised priority list before their shift; all 6 work lates.” |
| “Employees want more training.” | “4 of 9 new hires asked for a second shadow shift; their trainers were not interviewed.” |
Disagreement is part of the finding
A minority view is not noise to round away. "2 disagree" may be the two people at the site that does things differently, or the two who have been there longest. Read what they said. Sometimes the disagreement is the most useful finding in the study, because it points to a place where the problem does not occur.
What a small study can and cannot claim
| A small interview study can | It cannot |
|---|---|
| Describe how a process works and where people say it breaks | Estimate how often something happens across a large workforce |
| Surface explanations nobody had considered | Prove which explanation is the cause |
| Show where groups disagree | Show that a group's view is uniform |
| Point to the next question or test | Replace a survey as a comparable measure across everyone |
When the question needs a measure across a whole population, use a survey for that part and interviews for the why. The employee survey response-rate guide covers participation for the survey part. The qualitative people analytics guide covers how to analyse the interviews.
A fictional example: the second round
A fictional software company asks ten account managers why renewals slip past their dates. Seven describe waiting for legal review of contract changes; three describe customers delaying for budget reasons. All ten work in the UK office.
The team does not report "70% blame legal". It reports "7 of 10, all in the UK office", then asks six account managers in the US office the same question. Two of the six describe legal delays, and four describe a different approval step in finance. The finding changes from "legal causes slips" to "the approval that delays renewals differs by office", and that leads to a different action.
Where Lontra fits
Lontra reports findings the way this article recommends. Each finding shows how many people said it out of how many were asked, how many disagree, and what is still unknown, with every point linked back to the conversations behind it. Every quote is checked against what people actually said, and Lontra tells you when the evidence is not enough.
The Free plan includes ten interviews every 30 days, enough for a focused question. Paid plans add more interviews for questions that span more groups. Deep Study can go back to people whose answers left gaps, and to colleagues they mention.
It does not turn ten conversations into a statistic about your whole workforce, and you should not present its findings as one.
Enough to decide the next step
A small study rarely aims at a final answer. It aims at enough evidence, honestly reported, to decide the next step: act, test, ask a different group, or measure with a survey. Ten good conversations, reported as ten, can do that.
Frequently asked questions
- Is ten interviews enough for qualitative research?
- It can be for a focused question asked of one group that shares an experience. If experiences may differ by site, shift or role, include people from each group and report each separately.
- Should interview findings be reported as percentages?
- With small numbers, counts are clearer and more honest, such as 7 of 10. A percentage implies a precision and a representativeness that a small study does not have.
- What is theme saturation?
- The point at which new interviews in a group stop adding new themes. It is a useful rule of thumb, but it applies only to the group interviewed and does not show what an unasked group would say.