What’s a good sample size for qualitative research?

The standard in qualitative research is that it takes 12-13 responses to reach saturation. Meaning, whether you survey 13 or 130 people, the number of insights/themes you get is the same.
Want articles like this straight into your inbox?
Subscribe here

I consistently come across folks who are unfamiliar with sample size requirements for qualitative research -- assuming it needs stat significance like a quant test like an A/B test. 

Instead of stat significance the methodological principle used is 'saturation'. 

The standard in qualitative research is that it takes 12-13 responses to reach saturation -- meaning whether you survey 13 or 130 people, the number of insights/themes you get is the same. 

There are folks who debate the exact number of participants, but most in the scientific community agree it's below 20. 

A review of 23 peer-reviewed articles suggests that 9–17 participants can be sufficient to reach saturation, especially for studies with homogenous populations and narrowly defined objectives.

Hence our recommendation is to target ~15 people as a target sample size in your qualitative research.

What is data saturation?

Data saturation is the point at which new data no longer provides new insights into the research question. 

It’s when you have learned everything you can from the data and cannot find anything new. Data saturation is not about the numbers per se, but about the depth of the data (Burmeister & Aitken, 2012).

There is no one-size-fits-all answer to how many participants you need to reach data saturation. However, researchers agree on some general principles:

  • If you are not getting any new data, themes, or codes, then you have reached data saturation.
  • You should also be able to repeat the study and get the same results.

Some researchers have found that you can reach data saturation with as few as six participants  (Guest et al., 2006), but it depends on the population you are studying. 

More participants just means more costs

The vast majority of your target customer research should be qualitative. The point is to collect insights to drive demand, not big numbers to impress people.

Qualitative research with 15 people is a good investment because it yields the most findings at a lower cost. Running qualitative research studies with more than 15 people provides little additional benefit (you will hit saturation at around 15 people and identify 99% of insights) but costs quite a bit.

Spend that extra budget on more studies, not more participants.

Why qualitative research doesn’t need the same numbers as quantitative research

Qualitative research doesn't need the same numbers as quantitative research because it is focused on understanding the depth and complexity of people's experiences, rather than making generalizations about the general population. 

This type of understanding cannot be achieved by simply collecting data from a large number of people.

Just because 10 people in a 15-person study claim a strong interest in X does not mean that we can say that 66% of the overall population will have a similar preference.

Another thing is that qualitative research is often exploratory in nature.

This means that people conducting the research are not sure what they are going to find before they start collecting data. 

Qualitative research is often based on small samples of participants who are carefully selected to represent the group of people that the researcher is interested in studying.

This means that the researcher can be confident that their findings are relevant to the group they are studying, even if the sample size is small.

Sample sizes in user testing

Nielsen Norman Group recommends testing with 5 to 15 users to find most usability problems, as testing more people yields diminishing returns.

The math is explained in the chart below:

  • 0 users will find 0 problems. (Duh)
  • 5 users will be able to identify 85% of the usability problems
  • 10 users will find over 95% of issues
  • 15 users will identify over 99% of issues

The same principles can be applied to message testing as the key idea is the same: you’re trying to identify sources of friction. 

Know exactly what your buyers want and improve your messaging

Join 10,000+ other marketers and subscribe and get weekly insights on how to land more customers quicker with a better go-to-market machine.
You subscribed successfully.