As software product managers, we are chartered to unearth painpoints by interviewing customers/prospects. But how many do we have to talk to before we feel comfortable that we have talked to enough? How many is too many?
I have always used the following guidelines which I picked up from the great book “Customer Visits” by Edward McQuarrie (If you have not read this book, I highly recommend it. I consider it the bible on how to do effective customer visits). His research shows that interviews of
- 30 customers in the right market segment will identify 90% of the needs
- 20 would probably identify 80-85% of the needs
- 12 will identify 70-75% of the needs.
I typically start with 5-10 customers and see if I am already seeing convergence and then decide if I need to probe any specific areas even more by talking to more people. Make sure that you have randomly selected these customers from the target market segment and that you are not talking to same 5-10 everytime you need to do customer visits.
If you enjoyed this post, please consider leaving a comment or subscribing to the feed to receive future articles delivered to your feed reader.
3 thoughts on “How many customers does a software product manager need to visit/interview?”
I didn’t read the book. But by simple calculation using some of sample size calculator online, I would imagine what Edward uses is following assumption
– margin of error (if 65% say yes, still could be 50-50 split) = 15%
– population (total customers in that market segment) = 2000
So my interpretation is that you will have 70% confidence to say that those 12 customers reflects the population. Even that if you can get 2/3 of them (9) saying they want it, you still have 30% of chance that most of your targeted customers don’t need it.
10 is what I consider minimum but the numbers you post from the book are accurate in my experience. Its diminishing returns the more you do.
I also find that face-to-face is more valuable than videoconference, videoconference is more valuable than phone and email is useless (almost).
How do you feel online surveys fit into this picture?