Same method. Every level of the org.

I've never had a preferred domain. I've had a preferred kind of problem: rooms where the rules aren't written yet, and somebody has to write them.

I've worked that problem in CRM data, in real estate, and for six years in regulated healthcare, where the stakes and the expertise run deepest. At UPMC Enterprises, that meant setting venture-investment theses and advising as a board observer from the top of the org, and, on a venture I architected there, doing the hands-on product work myself.

Either way, the method is the same: learn the domain from first principles, find the real constraint, and build against it.