FUN FACT: if you start talking to commercial brokers, they'll whip out all the demand data you could possibly want. It all exists as the datasets veen works with for a living, and realtors can pay like $200/mo to be able to run whatever reports they want.

Barring that, you can do your own "winging it" version, which is what we started with: I looked at what characteristics a "home birth/birth center birth" consumer was likely to have, and then what other industries sought out that consumer. I started with "yoga studios" but had too large a dataset (and yoga instructors, I prejudicially assumed, didn't do that much market research). I then went with "natural food stores" and discovered that I could very easily correlate successful birth practices with proximity to natural food stores. At that point all I had to do was go "how many birth centers are there around this Whole Foods" to know whether or not the market was saturated or not. As a null hypothesis I pondered what demographic was diametrically opposed to my young, upwardly-mobile hippies and decided on "audiologists and hearing aid centers" for a category and "Five Guys Burgers and Fries" for my large corporation that does market research. My null hypothesis was null as fuck. No correlation whatsoever.

Veen then did some fancy-pants python-scripting isochrone-mapping demographic-parsing census data voodoo that said pretty much what I already knew, but with much greater detail. Six years later and our deliberately-conservative estimates have been blown out of the fucking water. The research was sound, the data was good, and we won.