Sample size of 30 in Market Research

If you ever took statistics, there’s one number you should remember. 30. Why is that? (I actually don’t know… do you?)

Here’s a funny idea – doing anything meaningful without some kind of research.

There is no ideation without research. As a UX designer, technology consultant, and digital marketer for two decades, it occurs to me that good ideas come ONLY from research.

So it’s not just for the marketing department, which is an important thing to note (there is no such a thing as “horizontal specialization”, just knowledge accumulated; there is only vertical specialization – agree or disagree?).

But how to research? Especially if the goal isn’t so much groundbreaking science as it is the generation of good ideas in business, however you define that. How to structure research?

As you can see, by the way, someone is making me think about market research in a new way. Someone whose newsletter I subscribe to, in which he talks a lot about things like probability samples, convenience samples, and census.

For years meanwhile  I have asked this question: why is 30 supposedly an ideal sample size? I have faith that it is, to an extent, because someone I respect (a university physics professor with a math background) told me so.

“To get a meaningful sample size of anything, you need 30 samples”, this professor would say.

This applied to, for example, Salmon run data provided by “the core” (the US Army Core of Engineers), whose responsibility it is to manage dams and the salmon populations which traverse them. So to draw a meaningful conclusion as to, say, the 40-year trends of salmon populations, you would need to get data from 30 dams.

This seemed beyond absurd and still does. What if there are a total of 30 dams in existence vs 3 million? Well now I have partial answers: if the total population is 30 and the sample is 30, then your research takes on the nature of a census.

Otherwise, however, it is either a probabilistic sample or a convenience sample, a distinction determined not by sample size relative to population but by approach.

That still doesn’t explain why 30 is such a magic number. If you could help enlighten me with explanations (yours or those of others), I would appreciate it. Not afraid to read in depth on this either, so feel free to suggest your favorite statistics novel, especially if it features stormy and complex romantic protagonists.

Again, though, I’m not looking to cure cancer or to create more irrefutable proof that the governments of the world are destroying the planet we inhabit. I just want to cultivate insight into the people I help as a marketing consultant: tech entrepreneurs, with both SaaS and solutions revenue, with an international business scope.

I’m reminded (in a sort of all roads lead to Rome way) of Ogilvy on Advertising, who said, “talk to your customer”. What he meant was, statistically hygienic and large scale market research is nice and all, but you learn more talking to 5 real people. Not sure how that qualifies as science, but I’m sure no marketer  has ever (a) achieved similar results or (b) done so with the literary panache of an culture critic.

Anyway, I suspect he used some blend of science and artistry.

As I try to solve the sample size of 30 mystery, I’m looking to rediscover that blend for the digital era.

I haven’t written in a while so please overlook typos – also gaping flaws in logic, fluency, and style. Thank you,

Rowan

July 8 2019