I feel like sites like Glassdoor became a victim of their own success. It became the salary site for years, so naturally it accumulated data going back years. Last I checked, their salary numbers are significantly lower than current industry comp since it has been going up over the past few years and it seems like they don't drop old data from the averages.
Can anybody vouch for the accuracy of the data? A friend of mine works for one of these companies, gets (or claims to get) excellent reviews, and makes nowhere close the "career trajectory" numbers.
BTW, salary data for H1B applicants is public and could be used to validate the base salary numbers, to some extent. Although it is not a given that H1B workers make the same as non-H1B. And then of course there is the stock, which is a mystery.
My own preference would be to value stock grants for total comp comparison purposes by the expected value of the number of vesting events, or, equivalently, to impose a discount rate on the future grants based on the turnover rate in the job category. I've never seen that applied to total comp comparisons; presumably it would be a moderate value somewhere between the lower bound and upper bound.