We’ve been asked that question so many times that I was compelled to write about it.
The real answer is, “it depends.”
But “it depends” to our clients is as useful as a sliding glass door in a submarine. We must give them an answer.
So, here are the variables we consider when providing a research approach plus insights gleaned from a dynamic discussion paper written by The National Centre for Research Methods. The report was written to help guide graduate students on the number of qualitative interviews they would need to do for a dissertation.
This report included the opinions of 19 academic research experts. A good number of them also concluded that “it depends.” But they also raised some excellent points to consider.
The paper indicated that the number of qualitative interviews in academia hovers between 20 and 60 interviews with 30 being the average. Still, others indicated 20 for an M.A. thesis and 50 for a Ph.D. dissertation.
For Market Research
But what about applied consumer or B2B qualitative research? Here are some variables to consider when determining the actual number:
- Research objectives. See below.
- The complexity of the topic. For example, if you’re trying to understand how the process of photosynthesis works in trees, you don’t need to talk to a hundred trees. You can talk to a handful of trees and understand this process, even though the process can differ from tree to tree.
- The size of your population/interview pool. I once conducted a focus group among supercomputer users and at the time there may have been a total of 20-30 users of such equipment in the US. But we were able to recruit 6 scientists in Southern California for one focus group.
- Theoretical Data Saturation, meaning that new interviews no longer provide substantially different information or insights.  That is, if you continue to talk to a certain number of similar people, you are likely to begin to hear the same thing.
- Homogeneity of your sample.
To us, homogeneity is key.
To illustrate, here’s a simple example. Let’s suppose your client wants to market to people named Karen. Yes, we know. Karen’s are extremely difficult, but then this is why the client wants to interview them.
Ideally, you could talk to 20 or 30 Karens, but is that too much? The question then becomes how many Karens do you need to talk to before you start to experience theoretical data saturation?
Here’s what we think, and it is backed up by the research and our experience. If you talk to 12 Karens there is a pretty good chance that Karens 13-20 or even 30 will begin to say the same thing, as we have discussed earlier. So, in this example, 12 Karen’s would suffice and 30 is likely too much, wasting the client’s budget, time, and money.
But now suppose the client wants to also market to Bobs. Now the sample becomes heterogeneous, and tradeoffs need to be made.
If the client has budget for 30, the problem can be easily solved by conducting 24 total interviews, 12 Karens and 12 Bobs.
But what if the client has budget and timing for only 12 interviews?
You can likely get away with six interviews among Karens and six among Bobs, but you may not get to theoretical data saturation or even begin to see patterns develop. But it may be better than no investigation at all.
Often, we will conduct a small pilot interview of three to five interviews and determine if more are warranted based on early results.
But remember, qualitative research is based on depth of information and not statistical reliability or predictability.