I was once pretty bullish on Glassdoor conceptually, potentially even around the time that The New Yorker profiled them, which I think was January 2018. Over the intervening years, I’ve soured on Glassdoor, for a couple of different reasons:
- The questionable nature of the reviews (more in a second).
- Their job postings have some of the most-insulting salaries of any job board, and that’s a very deep well of nothingness.
- I met a dude at a trade show in Chicago who told me his firm was charging some brands $15,000/month to “game” their Glassdoor reviews.
- I actually went to a conference at Glassdoor HQ once and it was raining + it was also “bring your dog to work” day, so the place smelled like 100 wet dogs.
Now we have Stanford University swinging for the fences on some Glassdoor-driven data project to determine what “culture” means in an organization. Essentially, they took 500,000 Glassdoor reviews, used some fancy AI/ML shit, and sought to “define” culture. That’s admittedly a better, more data-driven approach to “culture” than what most companies do, which is to build a children’s diorama of a mission statement out of toothpicks and lies, but it’s still very flawed.
See, the biggest problem with using Glassdoor “data” is that we have no concept of how real it is. Here are the essential issues:
- As noted above, some companies “game” their reviews.
- Very negative reviews are often from people who got terminated or otherwise embarrassed by the company.
- Very positive reviews are from sycophants, or they were offered a gift card to write a review, or some internal campaign happened.
- It’s hard to find a lot of nuance in reviews.
- In June 2014, I was struggling in Minneapolis and needed a full-time job. I got one in Texas, which is the initial reason I ended up down here. As I was negotiating and talking to my eventual-boss in Texas, I looked at Glassdoor reviews for the company, which were mostly horrible. I sent her an email noting that and she responded “What’s Glassdoor?” She made about $126,000/year and was pretty active on social media / “up on things.” She had no clue what Glassdoor was. I bet a lot of exec-types don’t either. As a result, a lot of the reviews are mostly just pointless venting or pointless ass-kissing.
When you combine all of the above, I don’t really understand how pointing AI at the site and scouring millions of textual phrases really helps you that much.
Plus: if you want to define culture, it’s actually not that hard. Does your team have your back? Are there good processes in place? Some semblance of work-life balance? Does the head of the silo know your name and what you do? If most of these things are being met, you have a “good culture.” If they’re not being met, you don’t.
Did I need AI and 500,000 cooked reviews to determine that? Not really.
Takes?
It’s cool that you asked your future boss about Glassdoor. Even if Glassdoor isn’t real data, asking about it was a real interview question–not some of the dog-and-pony-show ones.
These two statements are not far off:
> Very negative reviews are often from people who got terminated or otherwise embarrassed by the company. (although, it would be interesting to find out what the actual percentage is for “often”)
> Very positive reviews are from sycophants, or they were offered a gift card to write a review, or some internal campaign happened. (this makes it sound like all or even almost all positive reviews are from “sycophants” which is not necessarily true)
Two observations that I’d also make regarding Glassdoor rating are:
* the distribution within individual company’s ratings (LOTS of 2s, 3s, & 4s, vs. 1s & 5s at some companies, which goes against the “only people who hate or love the company leave ratings)
* the significant variance in scores across companies (Do some companies actively work the system and/or prod their employees towards Glassdoor? Absolutely. Do many of the companies with higher ratings on Glassdoor actually have better work environments and happier employees than companies with abysmal ratings? Yes again.