The volume of the data is relevant as it sets the degree of confidence we will have when making decisions relying on this information.
I define the confidence that I have in the observations I am gathering regarding previous knowledge and evidence about the topic too, so I can say that the volume of the data that I handle is not only the one from the project itself but also implies previous learnings.
Having said this, the volume of data needed to be collected in a project strongly relates to its goals.
My experience is mostly with generative studies, where the most recommendable is to start with small samples and recruit more participants when needed.
- A small sample would be five people. Comparative studies with more than one segment would be five people per segment.
- For generative studies, the volume of qualitative data is defined by information saturation. Once a pattern is discovered, I recommend validating it with a quantitative study such as a survey, an A/B test or tracking usage metrics in Analytics.
I know the confidence interval calculation to define the size of quantitative samples. I used it a while ago, but not in the last few years due to the strong dependency on the data that the orgs had from their users and customers and how suitable (and available) the CRM data (or similar database when available) is for UX research screening. What I do instead is
- to set a reliability degree for the data set that we are going to work with
- to work with some raw market numbers and establish the representativity of the sample segments
- to plan further studies when needed.