There’s a dashboard out there that knows when you’re going to quit your job.
It feels like something out of Minority Report. It’s definitely a plotline in the William Gibson novel Count Zero. But it’s not nearly as nefarious as it sounds, assures Kamran Niazi, a consultant at Deloitte. The sort of data a company would need to make such predictions? They probably already have it. The problem is, most companies just aren’t using that data in the right way.
That, there, is the human resources pitch for predictive analytics—linking myriad disparate data sources in an easily digestible software dashboard to help managers better figure out which employees are most at risk of leaving, and then using that knowledge to keep them on board. The reason, says Niazi, “is you want people to stay and grow because of the cost of losing people is huge.”
According to a 2014 Deloitte report, while 14% of companies regularly use data to make decisions about talent and human resources, only 4% are using predictive analytics to do so. Naturally, Deloitte has a predictive analytics solution of its own. But other companies do as well. Hewlett Packard, in 2011, piloted a predictive analytics program using factors such as salary and promotions to rank employees more likely to leave. SAP has its own solution it calls Infinite Insight, while Oracle has the rather frightening sounding Oracle Fusion Workforce Predictions.
There are also smaller upstarts, such as Workday, which jumped into the fray this fall with a product called Insight Applications.
It works like this: Some fancy software, hosted in the cloud, or perhaps on site, digests signals and inputs about a company’s employees–things such as an employee’s start date, time spent at the company, performance reviews, managerial structure, salary and number of promotions. This data is otherwise stored in separate places, and doesn’t mean much on its own. But the idea is that, by taking all of these factors, all of this data, and crunching, collating and linking it together, companies can predict not only which individual employees might be at risk of leaving, but wider trends across geographic regions or specific departments. Maybe high-performers in IT are quitting, and no one knows why. Or a company can’t keep its female executives for more than a few years.
“Customers are interested in data but they need more guidance, they need more help,” said Dan Beck, Workday’s vice president of technology products, in an interview. “I think we’ll look back in, like, not a lot of years—two or three years, not 10 years—and say, ‘Oh my gosh, people made hiring decisions without robust info about whether they’d be a fit for their organization.”
Of course, those trends won’t be the same for every company. There’s not some overarching rule that says “people in sales are likely to leave if they don’t get a promotion within three years.” The results are more personalized than that. You could focus only on a global company’s New York office, for example, where the attrition rate is higher for that location than the company as a whole. You might see that high risk employees all have three signals in common: the number of hours employees are working (a lot), the number of vacation hours being taken (not many), and the performance rating of those employees’ managers (not great).
If you zoomed in even further, you might so those risks change depending on position—executives versus middle management, say—or by ethnicity or gender, if that data is being collected too.
Perhaps unsurprisingly, the more data a company has, the better the insights will be. At Workday, Beck says one client showed up with 25 years of history on their 60,000-plus employees. “They gave us 65 inputs about their people, and we were able to say here are the 16 that are actually predictive to attrition, and here are the six that matter most,” Beck explains.
Where all of this gets really interesting is when external factors start to come into play. Deloitte, for example, takes into account local economic trends, depending on where a company’s various offices are located. Workday says it scours the internet for new job postings that match the skill sets of employees, so that clients can see who and what is theoretically in high demand.
And, of course, there’s that increasingly dystopian quantified future we’re hurtling towards, where everything from your wearable activity tracker’s data to how often you eat at the corporate cafeteria could possibly be used by your employer to build a more complete picture of you. Will these dashboards start to factor in risks like, say, web browsing activity? Attachments in email outboxes named “resume?” Whether an employee makes use of corporate perks, such as on-site gyms or cafeterias, or even FitBit activity data?
Niazi, of course, demurs. No one wants to go on record saying, “Of course we’re going to factor how long you spend in the bathroom each day” as a risk factor, an indicator of your work ethic. But he did agree that, as these dashboards and models continue to mature, so too will the sorts of data they can digest—be it data from inside or outside work, adhering to best privacy practices, of course.
For the moment, however, Beck says that companies just need to start leveraging the data they already have.
“It’s like you have a goldmine in your backyard and you never bothered to dig. We’re going to start there. And there’s a real long range I think to deliver there,” Beck explains. “This is not an absence of data, it’s an absence of knowing what to do with it.”