[tdumitrescu]

I've always found the reviews much more grim than the reality of places where I've worked. No surprise since it's a venue for departing and departed employees to vent (and then for HR teams to try to counteract by encouraging embarrassing good reviews from current employees), but in either case I've always found myself laughing at reviews of my workplaces, it's like "what company are these people actually working for? Because I'm pretty sure I don't go into a toxic hellscape of an office every day."

[cookiecaper]

Individual Glassdoor reviews aren't worth much because you never know the context or background of the author. Sure, it's possible that there is no mentorship and the place is cutthroat -- but it's also possible that reviewer was dead weight, actively rejected offered mentorship, and finally got let go after months of trying to work it out. Impossible to know in any specific instance, unless you already know the author.

It's really about the aggregation of reviews. If everything is fine until 6 months ago, and then there is a steady periodic stream of negative reviews referencing a management change, then you can pretty well-assured there was a legitimate shift in the organization that made things worse. If it's just a trickle of some positive (HR lackeys posting fake reviews and/or current employees trying to reassure themselves about how they work at the coolest place in town) and some negative (often bitter people who've been laid off or dismissed trying to project blame onto the company), then you just have to look for the consistent elements and take what you can from them. No easy formula.

[SemiEarlyGoogle]

I like to look at what I think are more authentic, indirect indicators, even though very imperfect, too.

For startups specifically, I used to analyze the startups that VCs tweeted the most. Uber dominated for a long time. And Zenefits was tops too, until they dropped fast in mentions a few weeks before their turn of fortune. Yeah, some tea leaf reading and eye of newt stuff, but interesting nonetheless and some job seekers found it helpful.

Your point about departing/departed employees and hellscape made me think about how the recent negative news about Google contrasts with one indicator (again, imperfect) of how happy people are to be Nooglers and Googlers -- their public exultation in Instagram posts: https://gxjam.com/ig-gallery/noogler-instagrams-by-googlers-...

Edit: Of course, neither of the above approaches are as broadly applicable as Glassdoor's approach

[spynxic]

This overall problem might be solved by offering periodic reporting where each subsequent report by a user is a bit more credible.

[ythn]

Reminds me of ex-[fill in religion] online forums. It basically filters on people that hate [fill in religion] and so it portrays only an extremely negatively-biased ("toxic hellscape") view of [fill in religion].