I concur wholeheartedly. Glassdoor is the absolute worst representation of salary ranges available. One could speculate as to why, but anecdotally, I've never received an offer that fell below their "reported" top-end.
AngelList is similarly bad for top-end comp representation, with the added twist of permanent anchoring: if you apply to a company that's listed an unreasonably low salary range, even if you know they'll (logically have to) pay more, they'll know that you can be bought cheap.
Why would anyone do that?
More commonly though I think people just aren't aware of the market situation, and comp ranges for specific jobs at specific companies.
Knowing the lay of the compensation land in your area is pretty critical to negotiating well - from both sides of the table. And this goes beyond generalities like "market salary" but to specific recent offers made for specific jobs at specific companies.
Honing the skill to determine whether or not a company is going to be able to pay you what you're worth is hinged on a few (apparently tricky) soft-skill-based assumptions:
- You can assess the company (quality, tier of engineer, etc.)
- You can assess your own market value, apparent worth accurately.
I like to give companies that I believe won't pay correctly the benefit of the doubt: I give a blowout number, xx% more than I believe I am worth. If I'm not met with resistance: I'm wrong and it hurts.
But I'm sticking with the strategy.
Ultimately the jobs where, in a vacuum and asked to value the job themselves, they come up with a number a sad fraction of your current comp, there was simply no scenario in which you would have been happy with comp at that place. It is simply the wrong job/company, and a change in negotiating tactic would not have fixed it.
The lowball initial offer ruffles some feathers and produces bad feelings, but ultimately does not change the outcome of the game.
Where withholding your current comp matters is where you're looking to make a step up - either a position with more seniority/responsibility, or simply at a company that pays people more. In these cases you have some knowledge that the upper end of the envelope is higher than your current comp, and so you don't want to anchor the negotiation low right off the bat.
Knowledge of how much companies pay for what positions is invaluable in negotiation - personally I wouldn't even bother interviewing at places where I know for certain the comp won't hit my expectations.
Side note: Glassdoor is incredibly bad for this kind of intelligence. Their medians lean low and the upper end of the envelope is in my experience almost entirely unrepresented. If you go in assuming the comp envelope is what Glassdoor says it is, you will lowball yourself severely.