Personas is a business fad I'm skeptical about. They can often be little more than stereotypes, and they often make the "high dimensional orange" mistake.

You imagine your persona has various attributes, of course. Like political preference, religiousness, age, income, melanin level, love of dark chocolate, love of Grateful Dead, you name it. And a lot of these are probably strongly correlated.

But if you normalize these attributes according to how much they vary, say on a scale from -1 to 1, then the average person is 0 on everything. But that person is vanishingly unlikely. So is the person who "follows all the stereotypes" and is a straight 1. Most people will lie on a "shell" at a specific distance from 0. That's the high dimensional orange: most of its volume is in the rind.

Your users/customers will feel pigeonholed when you assume that just because you like Grateful Dead, you'll like Santana. Or that just because you voted for Trump, you must love Prager U. Or that just because you've got high melanin, you like basketball. You get the picture, it's very easy to start annoying people with your "persona" assumptions.