Sample Size


Hello everyone, please clarify me with this doubt…

What should be the minimum sample number of the users to perform a user research activity of a very small size? Say 25 people.



Calling all @Researchers



I am not totally clear on what you are asking. Ideally, you want to Interview 8 to 12 people per target audience, or until you see patterns in behavior.

If you are usability testing it is 5 to 7 people.

Hope that helps.


Thank you for your response, I’ll reframe the query this way…

Suppose the target audience is 25 people, can I pick the sample of 3-4 people?

The research activities opted after learning user requirements are to create personas, scenarios, closed cart sort followed by AI and interactive prototyping.

Now the question, can I continue with these above mentioned research activities considering 3-4 people?



I in my experience to create persona creation liked to interview 4-5 people but i have quite lengthy interview (1h) - 3 is a good number already. I also like to look into job adds, that describe the role of the persona i am researching about.
I would say always the number of people depend on what you want to achieve! also how much time and resources do you have?
For statistical purposes there is a rule relating the number of query to the number of participant - like 4 participant per item. like if you create a scale and you got 10 items you want to validate - you’ll need 40 people…
Again it depends - what are you trying to achieve.


Hi Manu
Are you saying that the end users are only 25 people in total?
Generally for qualitative user research, it’s good to start off with about 5-7 people per group/cohort. I’m an advocate for smaller sample sizes, but more frequent research rounds. Keep researching until you’re not learning many more new things about your users.


Hello Ruth / Marion,

Yes, 25 people in total and doing a qualitative and quantitative research.

By doing telephonic interviews and user profiles so far with 4 users I have started to infer similar data outputs as the user roles played are of two types.

I was skeptic, if only 4 users was the right choice to perform the activity or to try for more in numbers.



I like this as a general rule of thumb for when to size up, i.e. when the “insights curve” starts to level off, that’s when you need to ask more people and/or ask different questions.


Hi @manu2cal,

Are there only 25 people total who will be using the app or service? You can do all of those activities, but you don’t want to be using the same people for each research activity. Where did the number 25 come from?



Hello @jdebari,

Yes, the total number of people using the app are 25 in number, as I learnt during the user requirement convenes with the stakeholders. (It’s an enterprise application, for a niche audience)

As we have a time crunch, I planned to take on a minimum of 4 users (the same ones) for each of the research activities as they were difficult to reach out plus in a foreign location.

I agree to your point, if different users are considered for different research activities we can glean more information but not happening as planned.

Kindly suggest me if things can still be improvised as you see with your experience.



There’s a lot of good advice in this thread re: Sample size and learning curve for qualitative research.

For quantiative research, if you can get it with a low level of lift on the part of the users (e.g. Google Analytics which happens in the background automatically), get as big a sample size as possible.



I’m curious about this @jdebari Can you unpack the benefit of having a research cohort of 8 split in two and doing two separate sets of tasks vs. having the whole cohort of 8 do the same tasks?

My gut instinct is the split approach would lead to erroneous data.


Hi @anne_dougherty,

If you use the same people for multiple different types of research activities, they become biased. So if you are doing in-depth discovery interviews, you want to use different people for usability testing because the people you already interviewed will have a mental model based on what they told you in the interviews and their own expectations. This article has some related points,


Hi @manu2cal,
If you can, try to at least fo the usability testing with different people than your first activities. It will help validate your design decisions better with an unbiased audience.


Hi @jdebari,

Makes sense. I think I may have been interpreting “research activities” too narrowly when I asked my question.

Thank you also for the link. :slight_smile: