I think it's a false statement to say someone less expensive will run things less well. I've been at a number of not-for-profits, and that does not match my experience at all.
Compensation should be set at a level where people don't need to worry about money, with a reasonable but basic standard of living. In the Bay, unfortunately, that's around $200k. On the other hand, compensation should not be set at a level where people are there for the money. You want people to be there because they are passionate and care.
$200 will bring people just as competent as $800k, but (1) your burn rate is lower (2) people are there for the mission (3) you can ask for donations in good faith.
Depending on part of the country, you can step that down significantly, in turn.
If you do want to invest extra money on people, a better place to spend that is stability and benefits (in the way universities do). I'd gladly take a job for $200k with a lifetime guarantee of doing meaningful work over one at $800k without that guarantee.
I actually think a lot of the bad decisions by Khan Academy are related to misaligned incentive structures. Khan tries to keep a pretty deep moat protecting business models. The platform isn't open source. Partnerships are hard to come by. Research partnerships are exceptionally difficult (Khan data is a proprietary resources). There's a cult personality. Etc.