24 thoughts on “Data Deluge and a Startup's Glassdoor”

  1. i too like the concept behind the glassdoor.com. what concerns me is how accurate the data (reviews & salaries) they are capturing will be. what’s to stop a company from supplying fake data just so that company looks good. or what’s to stop an upset employee from trying to take a company down by supplying slandering remarks? obviously the glassdoor.com has developed filters for these types of things, however, it’s going to happen & how much of the bad data that gets through could really hurt a company.

    similar to hotels.com. i wonder when i’m reading a hotel review whether it’s from an actual guest or if it’s been wrriten by someone that works at that hotel & is simply promoting their hotel.

  2. Om,

    You said it very well about the need to provide visitors and users with actionable information and knowledge they can actually use in their business or personal life.

    Understanding how your visitors can best use your site and the information contained in it – called use cases – is something that the big blogs appear to be sadly lacking. I believe it’s because you (the big blogs) fundamentally don’t view your websites in customer-focused terms. Instead, you are optimizing to maximize page views, which is at odds with efficient information gathering by your users, because your users want to click as few times as possible to get relevant and useful information.

    Here’s a few questions for you: Have you done a comprehensive user survey of all the existing aspects of your site and potential new functionality and content? Have you done that by use case?

    We’ve done that and the results were quite illuminating…and surprising.

  3. Om,

    You say there’s a huge opportunity for use of data. What is the single biggest category of information that one should look at on the web?

  4. what worries me most about glassdoor is NOT glassdoor itself, but companies looking for ways to make private data into money…with little incentive in place, throngs of disenfranchised employees will gladly toss up data and opinions and clutter sites like this with jaded commentary and distorted information (e.g. crappy employees get lower pay and lower bonuses, so numbers aren’t going to work consistently and will ruin averages because the N is too low in total)…

    happy employees want to stay put and move ahead and are far less likely to share information with sites such as glassdoor because they likely RESPECT their employers demand for confidentiality. glassdoor does not. companies like glassdoor are saying, “look, we know they told you not to discuss this publicly and we know that you can get fired for it, like the guy at google who wrote about it on his blog, but we’re gonna make you anonymous and give you a little treat too, so please tell, please violate your employer’s trust. thanks”

    the small 5 person or so team in place at glassdoor ‘checking data and opinions’ that roll in? come on…there’s no need. out of conservatism, just assume that most are lying or padding or have an axe to grind..

  5. @ pasan,

    I am having a tough time answering your question, mostly because it is too broad. You ask me for category of information – I need you to be more specific. Sorry but need help figuring that out.

  6. @ Don Jones,

    You bring up some good and valid points, and perhaps we should do a survey of our readers. However, to assume that we are optimizing for page-views is a tad presumptuous I think, especially if you have followed our site and the restraint we show when it comes to selecting the right kind of stories, especially for our audience. One thing I am very cognizant of – respect for our reader’s time.

  7. :::Robert was kind enough to run a little analysis, and indeed, despite the low input data, in the computer hardware sector there is “clearly a correlation between overall company rating, CEO approval rating and the price/earnings” ratio of the stock.:::

    Om immediately jumps to the conclusion that :::Loyal and happy employees work harder, make a company better and thus increase shareholder value.:::

    Although I’d agree with Om’s truism, I think without some sort of trend data on the three data points for each company it’s a little early to state definitively which way the correlation runs, especially with companies whose employees tend to have stock options as noticeable parts of their remuneration.

  8. Who’s not UNHAPPY about their salaries or working hours? Who does not feel that you are overworked and underpaid? Understanding where you stand on salaries is good but not the complete solution. We, at Leaders2020 believe that loyal and happy workers exist longterm in a company primarily because of the quality of their job and if the culture of the company fits with their values. The question then is how do you find companies whose culture matches with your values? That’s where our app, OfficeBook on Facebook comes to play. OfficeBook is our way to give back to millions of professionals who seek to work in a place where one is valued for who they are and what they truly believe in.


  9. What a great post and reminder about the idea of “actionable wisdom”. There’s a huge focus on quantity of data and materials both in the press/blog world and in corporations, with little enough qualitative balance. Anyone can rehash numbers or link to statistics, but it’s the connections between data points and analysis of data that’s important. Few people want to put in the extra brain power it takes to form useful conclusions. Didn’t we use to call that critical thinking back in school? Why does it seem to be missing in the post-graduation world?

  10. @ Thomas,

    I agree it is a smaller data set, but the conclusions do make common sense. how many companies do you know which have done well with a cadre of less loyal and unhappy employees. I think if people are happy, they will think of new ways. but I clearly think as Glassdoor gets more data, we get a better understanding of this all. The point of Glassdoor as an example of an entity that needs to think of their data in a large context.

  11. @ Mari,

    As someone who didn’t get around to be a post graduate, I don’t know what they teach in school. What I can tell you is that we really need to pause here and get some thinking in our work instead of chasing noise. news for the sake of news is not really valuable to anyone.

  12. “there is “clearly a correlation between overall company rating, CEO approval rating and the price/earnings” ratio of the stock.”

    Really? Eyeballing the numbers from the chart, the correlation coefficient for Company Rating and P/E is 0.19, which doesn’t sound like a very strong correlation to me. For CEO rating and P/E it’s 0.58, which is better, but still not (especially with a sample size of 8) all that compelling.

    And don’t even get me started on the form and design of the chart…

  13. @ DG Lewis,

    The thing is there is still a correlation between those factors and as those data sets bulk up the fidelity of the information will increase.

    That said, your focusing on the PE & the chart highlights exactly the problem I was trying to highlight – you are seeing a specific part of the data. I am saying lets use this data to paint a bigger, more clearer picture.

    “it’s not data that’s important but what you do with it” Perhaps i failed in making my case through my words, and that is why you focused on the chart. oh well, I shall try again sometime soon.

  14. Hi, Robert Hohman from glassdoor.com here.

    Thank you all so much for the interest in glassdoor. We are humbled by the community’s response to our site, it has truly been amazing.

    @Brian Kirk:

    We’ve taken data integrity extremely seriously – in fact our current run rates have us rejecting about 14% of content because it is either silly, too high/too low/bogus, jeopardizes anonymity, or appears to divulge trade secrets. We review every single piece of content by hand before it goes live, and to be honest, we’re pretty buried 🙂 But we’re catching up.

    As I shared with Om, the really exciting stuff is still to come. Right now we are very focused on data collection, but give us a couple of months and we will deliver to you some extremely interesting ways to dig into this data.

  15. @ Roberthohman,

    Glad to see you are taking the fidelity of the content even more seriously that you let on in our conversation. Well that’s what happens when we are bound by limits of time and tyranny of appointments.

  16. Om:

    Data deluge is a very interesting topic and the inundation done by streams coming out of FF, twitter, blogs, etc. are far from actionable. What we have taken as given that a majority of web data gets created for it’s search destiny (and maybe drive traffic from point A-to-B).

    Would love to see your separate take on this topic (the deluge, the players, the incumbents, the opportunities, etc.).


  17. Folks – Check out this website http://www.thesmartpro.org which is pretty much the same without the unwanted details. How many employees want to know the approval rating of the CEO before deciding to accept the offer?

    Smartpro is neat, it classifies the responses under categories that are important and relevant to employees. It’s free and not for profit!

    It was created by the employees and for employees and funded by the employees too and much before glassdoor was conceived.

    check it out at http://www.thesmartpro.org


Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.