How Many AI-Moderated Interviews Are Enough? (2026)
Qualitati Research Team · 2026-06-20 · 11 min read
Short answer: There is no fixed number of interviews that guarantees saturation. Classic guidance lands at 9–17 interviews for code saturation and 16–24 for meaning saturation in focused, homogeneous studies, but heterogeneous or multi-segment research routinely needs 20–40+. AI-moderated interviews change the economics — running hundreds becomes feasible — but saturation is still a property of your research question and sample diversity, not your tooling. Plan a minimum, monitor new themes per interview, and stop when fresh transcripts stop adding codes.
How many AI-moderated interviews are enough?
“How many interviews do I need?” is the most-asked question in qualitative research, and the rise of AI-moderated interviews has made it sharper, not simpler. When each interview cost a recruiter, a moderator, and an hour of someone’s time, the answer was shaped by budget as much as by method. AI moderation removes most of that marginal cost — so the real constraint shifts back to where it belonged all along: when have you actually reached data saturation?
This guide explains what saturation is, what the evidence says about sample size, how AI changes the calculus, and how to decide — with a stopping checklist you can apply to your next study.
Key takeaways
- Saturation is the stopping rule, not a magic number. It is the point where new interviews stop producing new themes or codes.
- Ranges, not guarantees. Code saturation often appears by 9–17 interviews; meaning saturation by 16–24; heterogeneous studies need 20–40+ (Hennink et al., 2017).
- The Guest 2006 “nine interviews” finding is widely over-generalized — it described one homogeneous sample answering one focused question (Guest, Bunce & Johnson, 2006).
- Sample size scales with five factors: aim breadth, sample specificity, theory use, dialogue quality, and analysis strategy (Malterud et al., 2016).
- AI changes the cost, not the principle. Running 300 interviews does not buy saturation if all 300 come from one narrow segment.
What data saturation actually means
Data saturation is the point at which additional interviews stop yielding new codes, themes, or understanding — you are hearing the same things in different words. It is the most common justification for a qualitative sample size, but it is frequently misused as a single threshold rather than a judgment made against your own data.
Researchers distinguish several kinds of saturation, and conflating them is the root of most sample-size confusion:
| Type of saturation | What stops appearing | Typical range (focused study) |
|---|
| Code saturation | New codes / topics | ~9–17 interviews |
| Meaning saturation | New understanding of each code | ~16–24 interviews |
| Theoretical saturation | New properties of an emerging theory | Often 2× data saturation |
The practical lesson: hitting code saturation (you have heard all the topics) is not the same as meaning saturation (you understand each topic deeply). Reporting “we reached saturation at 12” without saying which saturation is a red flag a careful reviewer will catch.
What the evidence says about sample size
The number most often cited — saturation at nine interviews — comes from Guest, Bunce, and Johnson’s 2006 study. But that study used a homogeneous sample answering a single focused question. Under almost any commercial or multi-segment research condition, nine is not adequate, and treating it as a universal floor is a methodological error.
More defensible is Malterud, Siersma, and Guassora’s “information power” model: the more information your sample holds relative to your aim, the fewer participants you need. Five dimensions drive it:
- Aim breadth — narrow aims saturate faster than broad, exploratory ones.
- Sample specificity — highly specific participants carry more relevant information per interview.
- Use of theory — an established theoretical lens reduces the data needed.
- Quality of dialogue — richer, well-probed interviews carry more information each.
- Analysis strategy — a deep single-case analysis needs fewer cases than a cross-case comparison.
A useful planning heuristic synthesized from this literature:
| Study type | Sample diversity | Planning range |
|---|
| Focused, single-audience | Homogeneous | 10–15 interviews |
| Moderate complexity / light segmentation | Mixed | 15–25 interviews |
| Multi-role or multi-segment | Heterogeneous | 20–40+ (across segments) |
Plan to a minimum from this table, then let saturation decide the rest. Segment matters as much as total: 30 interviews split across six personas may give you only five per persona — nowhere near saturation within any one of them.
How AI moderation changes the calculus
An AI-moderated interview is a structured conversation run by an AI interviewer that asks your questions, probes open-endedly, and adapts follow-ups — in text or voice, in parallel, at any hour. This breaks the historical link between sample size and cost, and that has three consequences for saturation:
- Larger samples become practical. Teams that once planned for 20 interviews can run 200 — useful for convincing data-minded stakeholders and for genuinely heterogeneous populations.
- Saturation can be monitored, not guessed. When transcripts and coding are automated, you can track new themes per batch and watch the curve flatten in near-real time, rather than declaring saturation retrospectively.
- The failure mode shifts. The risk is no longer “too few interviews” but “many interviews, narrow sample.” A thousand interviews from one demographic still leaves you blind to everyone else.
One caution: bigger is not automatically better. Past genuine saturation, additional interviews add cost, analysis load, and noise without adding insight. Scale to cover diversity and to satisfy your decision-makers’ evidentiary bar — not for its own sake.
Original asset: the Saturation Stopping Checklist
Use this checklist to decide whether to stop recruiting. Treat “yes” to all five as a defensible stopping point — and document your answers in your methods write-up.
| # | Check | Stop when… |
|---|
| 1 | New codes | The last 3–5 interviews produced no new codes or topics. |
| 2 | New meaning | Existing themes gained no new dimensions or counter-examples. |
| 3 | Segment coverage | Each planned segment — not just the total — has flattened. |
| 4 | Disconfirming cases | You have actively sought, and stopped finding, cases that challenge your themes. |
| 5 | Minimum met | You are at or above your pre-registered minimum for this study type. |
If you answer “no” to #3 or #4, you have not saturated — you have simply run out of one kind of participant. Recruit for the gap, not the count.
Where Qualitati fits
Qualitati is an AI user research platform for product, UX, and customer insights teams. Its AI-moderated interviews and focus groups make larger, more diverse samples practical, in text or voice across 10 languages — so you can saturate within segments instead of settling for a thin spread across them. ThemeLens, its AI thematic-analysis pipeline, maps codes to research questions across up to 100 transcripts at once and synthesizes themes with participant-anchored quotes — which is what lets you actually see the new-themes curve flatten rather than guess at it. The QDA Workspace supports inductive and deductive coding so a researcher confirms saturation against the codebook. Qualitati is an AI-native alternative to NVivo, Qualtrics, ATLAS.ti, and MAXQDA, and a transparent-pricing alternative to Outset.ai, Strella, Listen Labs, and User Interviews.
Limitations and trade-offs
Three honest caveats. First, saturation is a judgment, not a calculation — the ranges above are planning aids, and reasonable researchers disagree on when a theme is “fully” understood. Second, automated coding can create false confidence: if an AI under-splits codes, the new-themes curve flattens early and you may stop too soon, so a human should audit the codebook. Third, large AI-moderated samples shift the bottleneck to sampling quality — recruiting and screening now determine whether your numbers mean anything, and saturation cannot rescue a biased sample. For methodology-sensitive claims, keep a researcher in the loop.
FAQ
How many qualitative interviews are enough?
There is no universal number. For a focused study with a homogeneous sample, 10–15 interviews often suffices; moderate-complexity studies need 15–25; heterogeneous or multi-segment studies routinely need 20–40 or more. The real answer is to stop when new interviews stop producing new themes within each segment.
What is data saturation in qualitative research?
Data saturation is the point at which additional interviews stop yielding new codes, themes, or understanding. It is the most common justification for a qualitative sample size, but it should be assessed against your own data rather than assumed at a fixed threshold.
Is the “nine interviews” rule reliable?
No. The widely cited finding that saturation occurs at nine interviews came from a homogeneous sample answering a single focused question. It is frequently over-generalized; most applied and commercial research with diverse participants needs substantially more.
Do AI-moderated interviews need more or fewer interviews?
The same saturation logic applies, but AI moderation lowers the cost of each interview, making larger and more diverse samples practical. This helps you saturate within segments and satisfy data-minded stakeholders — though running more interviews from a narrow sample does not produce saturation.
How do I know when to stop recruiting?
Stop when the last several interviews add no new codes, existing themes gain no new dimensions, every planned segment has flattened, you have stopped finding disconfirming cases, and you are at or above your pre-registered minimum. A stopping checklist makes the decision auditable.
Conclusion
“How many AI-moderated interviews are enough?” has the same honest answer it always did: enough to reach saturation for your question and your sample — usually a planned minimum somewhere between 10 and 40, then a few more to confirm the curve has flattened. What AI moderation changes is the cost of getting there and your ability to watch saturation happen in real time. Plan a minimum, monitor new themes per batch, cover every segment, and let the data tell you when to stop.
Start free with 30 credits, no credit card required, and run an AI-moderated interview or focus group with built-in thematic analysis. Create a free account, view transparent pricing, or read the AI-moderated interview pilot checklist.