Personally, I love the high stability of a broad consumer base. However, you can architect a successful business either way.
Apple charges a high price relative to its cohort but not its customers' aggregate spending - recent times see it it pushing towards r-strategist behaviour (lots of lower paying customers, e.g. iPad).
If Apple were pursuing a k-strategist mentality (few, high paying customers) I would not have to wait an hour to get serviced in New York, I'd have small batches of product with bleeding edge technology, and I'd have a personal support rep and probably custom app developer.
Every business has a different decision to make. It feels like this thread is thinking of it like maximizing an equation (FindingMaxRevenueGivenPriceSensitivies()). Instead, some products lend themselves to low cost support; sell those more cheaply. Some products require insurance-like support contracts -- those are going to be enterprisey. Some products will lend well to cheap inexpensive versions and high end enterprise versions, without changing the product (the value prop is really high to a business, esp if it's a service that can automate out the cost of a person's salary).
This thread and the obsession with A/B testing result from a hyper focus, like you said, on finding local maxima. I think it's often useful to take a step back and find the high level factors and built a pricing model around that.
(Of course, nothing excuses you from talking to customers.)