Mining for Quitters

  • Comments posted to this topic are about the item Mining for Quitters

  • Seems like a reasonable use of BI tech. Heuristic, self-correcting systems like that can work.

    I built one that predicted whether a customer would stay or leave, and it was accurate enough to be useful. Wasn't perfect, but no system ever is. I could see something like that having a use in employee retention.

    Would have to be a large number of employees before it would have a sample-base large enough to be useful, but Google probably qualifies there.

    Just so long as they don't use it to replace human judgement, but to supplement it, I think it's a good idea.

    - Gus "GSquared", RSVP, OODA, MAP, NMVP, FAQ, SAT, SQL, DNA, RNA, UOI, IOU, AM, PM, AD, BC, BCE, USA, UN, CF, ROFL, LOL, ETC
    Property of The Thread

    "Nobody knows the age of the human race, but everyone agrees it's old enough to know better." - Anon

  • Obviously not directed at Steve... this type of thing is a big steaming pile of hooie. Everyone thinks they can figure out humans with numbers and stats. Try the old fashioned method... TALK WITH PEOPLE!!!! IN PERSON!!!

    --Jeff Moden


    RBAR is pronounced "ree-bar" and is a "Modenism" for Row-By-Agonizing-Row.
    First step towards the paradigm shift of writing Set Based code:
    ________Stop thinking about what you want to do to a ROW... think, instead, of what you want to do to a COLUMN.

    Change is inevitable... Change for the better is not.


    Helpful Links:
    How to post code problems
    How to Post Performance Problems
    Create a Tally Function (fnTally)

  • I agree with you, Jeff, but who do you talk to? I've known plenty of people that were thinking of leaving, and wouldn't want to let their manager know. And we always find bad managers.

    Having something that helps you decide who to talk to is not a bad system, IMHO.

  • Steve Jones - Editor (6/9/2009)


    I agree with you, Jeff, but who do you talk to? I've known plenty of people that were thinking of leaving, and wouldn't want to let their manager know. And we always find bad managers.

    Having something that helps you decide who to talk to is not a bad system, IMHO.

    Agreed.... I've done it before... if I'm thinking of leaving because of a bad manager, it's too late. The talking process shouldn't start at the Developer level. The folks from the Ivory Tower should come down off the hill and talk to the folks in the trenches. Maybe even beltloop a couple of them for the day just to see what it's like. I'm not talking immediate managers, either.

    --Jeff Moden


    RBAR is pronounced "ree-bar" and is a "Modenism" for Row-By-Agonizing-Row.
    First step towards the paradigm shift of writing Set Based code:
    ________Stop thinking about what you want to do to a ROW... think, instead, of what you want to do to a COLUMN.

    Change is inevitable... Change for the better is not.


    Helpful Links:
    How to post code problems
    How to Post Performance Problems
    Create a Tally Function (fnTally)

  • There is also the factor of "company loyalty" to consider. If an exec has been poached from another company, doesn't that suggest the exec is less loyal? The exec is only interested in the next best offer to come along. Surely these people are the most likely to leave? I don't need complex heuristics to work this out 😎

    Andy

  • Capability to predict if an employee may be wanting to quit is a marvellous technology. It is somthing that will not be applicatble to all organizations / scenarios and may not be welcome by all people. So, this technology may grow only in committed organizations and verticals.

    A support center organization owner in India may be thrilled with this news as it can help reduce the attrition levels in the organization. So that is a target industry for the focus and development of this technology.

  • vinodk (6/10/2009)


    Capability to predict if an employee may be wanting to quit is a marvellous technology.

    My conviction is that this so called technology is all crap, they think they got something but I am sure time will disprove their claims. They did more predictions using it, one of them is that the economic crisis is going to be over by the end of the summer. How did they conclude that?

    There are less searches done by individuals on the topic

    Yeah that's right...talk about simpletons...as if there couldn't be any other reason for a decline. They did not even bother to put in verified facts and concluded this based on just search behavior! They assume that the peoples "media manipulated" feelings are correct and are thus a strong predictive indicator. Also the idea that if enough people believe something, it will become true automatically...now lets all think we can fly...we don't need airplanes and cars anymore then. BTW...how did we get into the mess in the first place...its not as if we all started looking for the word financial crisis.

    It's a "human assumption" based technology...that kind of sums it up IMO.

    And don't forget reverse usage as in pay cuts to those least likely to leave. Or what about a company using such a tool that can create new discriminating factors that reinforce themselves. Say you got 10 people working for you and your tool classifies the two that it thinks have the highest chance of leaving. The managers sees this and realizes that investing in those is more risky then on the others.

    This could then translate in a decision to not give those two some training that most of the others will get or to simply not put them forward for promotion. Clearly that will raise the real chance of those classified as most likely to leave...to leave. Thus the tools predictions **seem** right when looking back.

    But any feedback that does not take in account the decisions that were made based on its earlier predictions and where the "target" group was not split in a scientific manner (like a double blind test) will pollute its own model. Yes it seems like it has predictive power but in reality it does not have any at all. People making decisions are likely to make those predictions come true without them realizing it.

    The very idea is just fraud with problems.

  • Years ago I was a manager for a large auto glass manufacturer and installer. We had a an application that "examined" every claim for auto glass repair and replacement that would come in daily. The program would factor how many windshields were being replaced versus repaired (lower cost) on a daily basis across the entire United States.

    This meant that if you had your auto insurance with any of the companies we served, and you broke or cracked your windshield on any given day, a computer was deciding whether you would get a replacement or repair based on what costs were at that given moment. The application did not account for your safety when say, a repair was not the best idea - it just looked at numbers and dictated its recommendations to keep costs down, and insurance profits up.

    Later I learned that many insurance companies in various businesses do the same thing - including healthcare institutions.

    I could not disagree more with today's editorial asserting that an application be making decisions about human beings.

    We simply are not at that level with software applications where people's lives, welfare, health, even wages should be "analyzed" and dictated by computer programs. In fact, I consider the very idea to be borderline insane! Do you really want key decisions about you made by some calculation? Do you really trust software developers and designers THAT much? If so, I think you need your head examined.

    We have already seen computerized trading on Wall Street botch the numbers affecting millions. We have seen NASA send a probe to Mars where the computer told the lander to bury itself 200 meters deep into Mars surface. We have seen power grids go out because software "thought" it needed to. And these are just a few of the thousands of glitches software has caused.

    We are not ready to turn human decisions in any fashion over to computer software. Yes, I know this is already happening but it is insanity. Applications are written by humans and humans are flawed. Putting your health, welfare, income, and future in the hands of an application makes about as much sense as considering a Quija board a valuable diagnostic tool.

    I keep people on my staff by encouraging them, rewarding them, and guiding them along - I DONT want any application doing that for me!!! :w00t:

    There's no such thing as dumb questions, only poorly thought-out answers...
  • The dark side of this is they figure out who is likely to leave and fire them before they have the chance to quit:-P!

    Instead of wasting time trying to manage by numbers maybe they should look at why people leave and work on improving the overall climate so everyone is happy. Focusing on a few people who may be looking to leave may create a system that ignores the rest.

  • problem with this sort of system is it is just as likely to be used to identify who is NOT likely to leave and therefore you can get away without incentivising them. (I bet age would be a criteria leading us into a legal minefield)

    If HR have a system like this (it would be HR using it right?), its word would be taken as gospel (the computer says yes) and the all important human factor would be removed from the equation.

    ---------------------------------------------------------------------

  • I see this as a negative, kind of like police profiling, or lie detectors.

    Sure it's cast here as a 'good thing', but in the long run it can get pretty ugly. People perceived by the fallible system can be the first to be let go, less likely to be promoted, etc, and I think that's where the majority of action in this area will wind up.

    The biggest problem is that because the system is so fallible people will be miscategorized.

    ...

    -- FORTRAN manual for Xerox Computers --

  • What data would be used to determine potential quitters? What data could anyone possibly have that would be useful?

    What? Are they going to have some psychologist determine that blonde hair, thirty-two year-olds with expanding waistlines that just had their third child are more likely to leave?

    This is really bad. The more I hear and read about Google, I'm getting the idea that they may be using technology for bad purposes.

  • skjoldtc (6/10/20090

    This is really bad. The more I hear and read about Google, I'm getting the idea that they may be using technology for bad purposes.

    The problem is bigger than google. The problem is human nature. Mining and manipulation of information is inevitbably going to be expanded, it is simply too seductive to be ignored.

    When it's used by marketers, it's merely annoying. When it's used by governments and employers (and it WILL be, regardless of claims to the contrary) it is very threatening.

    ...

    -- FORTRAN manual for Xerox Computers --

  • Jeff Moden (6/9/2009)


    Obviously not directed at Steve... this type of thing is a big steaming pile of hooie. Everyone thinks they can figure out humans with numbers and stats. Try the old fashioned method... TALK WITH PEOPLE!!!! IN PERSON!!!

    It doesn't help predict individual behavior, and is obviously no substitute for human contact. BUT, in a company with thousands of employees, can top management, or even upper middle management, actually get to personally know every employee? Not realistic in groups larger than a few dozen people. And if the company is spread out over a large geographic area, with offices in remote locations away from the central management office, it becomes impossible even in smaller groups.

    Most people can manage about a dozen or two dozen personal relationships at a time. A few can manage a hundred or more, but that's very, very rare. Think about how many people you know, which has probably been thousands over a lifetime. Now think about how many of those you can actually keep track of how they're doing at work, their morale, job performance, ethical standards, current pay vs market norms, etc. That'll probably be a few dozen. How many of those is the data current in your mind, versus weeks, months, or years out-of-date?

    Given those factors, management needs some sort of ability to know who's at risk and who isn't. I think a system like this, used appropriately, makes sense for that.

    - Gus "GSquared", RSVP, OODA, MAP, NMVP, FAQ, SAT, SQL, DNA, RNA, UOI, IOU, AM, PM, AD, BC, BCE, USA, UN, CF, ROFL, LOL, ETC
    Property of The Thread

    "Nobody knows the age of the human race, but everyone agrees it's old enough to know better." - Anon

Viewing 15 posts - 1 through 15 (of 50 total)

You must be logged in to reply to this topic. Login to reply