With qualitative data, I never look for statistical significance. It is impossible with the sample sizes used and is not the “point” of qualitative evidence. However, that doesn’t mean that qualitative data isn’t prone to biases that need to be averted with good methodology.
For our first set of qualitative interviews, one of our goals was to avoid cherry-picking customers. We did our best to seek our and talk to a broad swath of customers. For example, we knew that some of our customers were from small companies and some were from bigger companies, some were e-commerce and others were travel. Instead of just talking to the first customers who responded to my request for an interview, I went out of my way to recruit a sampling from each group.
Reaching statistical significance for the cluster analysis (quantitative data) is a different, and equally interesting, question. Opinions on this vary and there is no rule of thumb. This paper has some great guidelines. One important factor to consider is that the more variables you want to add to the model (e.g. company size, number of years using product, NPS, job title, etc), the more responses you will need.