Corporate mind games

More and more companies are mining data about their employees to predict how they’ll behave in the future

Photograph by Liz Sullivan

A few weeks ago it was revealed that Google and the CIA—two organizations whose job it is to know what’s going on in the present—are working together to learn what will happen in the future. Through their respective investment arms—Google Ventures and In-Q-Tel—the geeks and the spooks each took a stake in a little-known Cambridge, Mass., company called Recorded Future, which bills itself as “the world’s first temporal analytics engine.” By scanning websites, news stories, blogs and Twitter pages for links between individuals, groups and incidents in the past, the company says it can apply “temporal reasoning” to predict events that haven’t happened yet.

The company is at the forefront of a fast-growing field known as predictive analytics, which uses algorithms that detect patterns and connections in life that wouldn’t otherwise be found.

The concept of businesses harvesting the past to try to predict the future isn’t new. But what has changed is the tsunami of information that gets churned out every day, serving as the fuel for these mathematical soothsayers. And while examples of police and spies using such techniques grab most of the attention, a growing number of companies are applying predictive analytics to the workplace to figure out which employees are at risk of leaving, who will work the hardest, and who is worth hiring in the first place.

It’s easy to understand what drew the CIA to Recorded Future. The company has already had some success forecasting troubling events. This past March, Israel warned that Hezbollah had obtained long-range Scud-like missiles, something Recorded Future had found evidence for a month earlier. It’s not the only one delving into the field of predictive analytics. Police in Memphis, Tenn., are using IBM technology to prevent crime before it occurs by predicting when and where it will happen.

In the past, companies used the technology when doing market research and to gauge the credit-worthiness of individuals. But large firms are beginning to use historical data as a crystal ball within their human resources departments, says Atanu Basu, CEO of Data­Info­Com, a research-focused predictive analysis firm in Austin, Texas. “Every large company already has access to this tremendous amount of data on their employees,” he says. “That can be used to make predictions about who will be successful and who is going to leave.”

Last year Google said it had developed a predictive tool capable of warning managers which employees will quit. An executive boasted that Google’s algorithm was so powerful, it lets the company “get inside people’s heads even before they know they might leave.” The U.S. Army already uses the information it gathers from recruits to predict which job would suit them best. And according to a recent report in HR Magazine, some companies are crunching the numbers to even predict when employees will call in sick.

The fact is, workers today shed personal information like skin cells that employers can collect.

The Web pages we surf at work, the emails we send, the answers we give during annual reviews, even the patterns of our vacations are all potential indicators of our actions, say experts. And as we expose more of our personal lives to the world through Facebook and Twitter, that task is only made easier.

Of course, there are risks any time a company thinks it’s got the future figured out, says Tom Davenport, a professor at Babson College and an expert in the field of analytics. Consider the damage wrought by the credit crisis in 2008. Wall Street suddenly discovered that its predictive models about the future of the economy—which assumed house prices could never go down—were completely wrong. It can be the same with individual workers. “The best predictor of how you will behave in the future is how you behaved in the past, however human beings sometimes change,” he says. “The person who resigns may not be the person you predicted would resign at all.”

But as companies fill their vaults with information about their employees, don’t be surprised if one day your boss looks at you funny. He may know you’re about to look for a different job, even if you don’t.




Browse

Corporate mind games

  1. The problem with these computer models is that people forget about the quality of the data being put in to them, and then overreact with the results. For example, stock markets go up in down as a result of changes in employment and inflation numbers, even when these changes are well within the statistical margin of error.

    As an other example, companies will often use predictive algorithms meant for other cultures. For example, asking a Canadian if he has ever been part of a sorority doesn't make sense, since sororities are not part of Canadian culture (and even less so in Quebec). Yet I have been asked that by two Quebec insurance companies when applying for a job.

    Also, people lie. Rogers used to ask prospective employees about what they thought about unions! Not exactly the best way to end up with honest employees.

  2. “Last year Google said it had developed a predictive tool capable of warning managers which employees will quit. An executive boasted that Google’s algorithm was so powerful, it lets the company “get inside people’s heads even before they know they might leave.””

    Boss: You’re going to quit, aren’t you?

    Employee: No, I’m not. I have a wife and kids.

    Boss: But you’re going to leave anyway. You just don’t know it yet.

    Employee: Then how would you know?

    Boss: Well, we’ve been looking at your employee forms, as well as your web browsing history and e-mails…

    Employee: I was browsing the web during my lunch break. As for the e-mails, I followed the template…

    Boss: I’m not finished. Then we went through your Facebook account, your Twitter feed…

    Employee: I don’t talk about work there!

    Boss: That might be so, but we also crunched the numbers on the socioeconomic data of the area within a kilometre of where you live, going back twenty years, we’ve consulted your family’s medical history through the health plan we provide — your wife could lose some weight, you know –, and we’ve gone through every detail you mentioned on your resumé…

    Employee: And…?

    Boss: And the political history of the federal and provincial ridings where you live, the age of your kids, the make of your car, your zodiac sign, the fact that your date of birth is divisible by seven, and that you’re unlikely to have gone on vacation in Andorra, Paraguay or Spitsbergen…

    Employee: And this all means that I’m going to quit soon?

    Boss: Pretty much, yeah.

    Employee: But how? Why?

    Boss: Because of those four little words you said during one of your lunch breaks: “I value my privacy”.

    Employee: Well, I do. What are you going to do about it? If this is for real, I’m writing to the Privacy Commissioner right when I’m getting home, but I’m definitely not going to quit. I need this job.

    Boss: Oh, I don’t doubt it, and we’re not about to fire you, even though the law says we could. We don’t fire employees without a good reason; we’re a compassionate company. But the janitor called in sick — we predicted that, too, considering he likes Elvis movies, train models and TV dinners –, so would you mind cleaning the toilets?….

    • Be a team player sport! Oh, and we're cutting your pay 23% to pay for my country club membership increase.

  3. I am really grateful to have the information from this blog.I liked the blog as it has been written,the information i got from here.This is a good information i got from here.I really liked it and this information is worth remembering.

  4. I am really grateful to have the information from this blog.I liked the blog as it has been written,the information i got from here.This is a good information i got from here.I really liked it and this information is worth remembering.

  5. Scary stuff.

    Specially when you realize that all those calculations are based “how you behaved in the past”

    It’s sort of like you are stigmatized for the rest of your life for whatever you did.

    Human beings change constantly.

  6. "the company says it can apply “temporal reasoning” to predict events that haven't happened yet." i wonder if it will predict the revolution?

  7. gotta watch Minority Report again !!

  8. disasters can't be predicted, no matter how hard we try
    people are not very predictable and how can computers process this when we are still trying to figure out how to communicate with each other
    computers don't take in emotion, thought processes, family dynamics, social dynamics, and all the other human baggage
    and it sounds like only people who are web based will be affected, there are still people out there who don't use technology
    We are always trying to figure out the patterns of life to make things PREDICTABLE so we can build safety nets.. And no matter how hard we try to figure out the OUTCOME through statistics, psychics or anything else, It has NEVER been done.. so save your money and just LET LIFE HAPPEN!!

  9. Predictions will never be exact. Predictions are simply educated guesses about what is likely to happen. Clearly PA is going through its early phase of the the Technology Hype cycle.

Sign in to comment.