Imagine one of your managers walks into their subordinate’s office and says, “Our data analysis predicts that you will soon get restless and think of leaving us, so we want to make you an offer that our data shows has retained others like you.” Would your employees welcome the offer, marveling at the value of your HR analytics? Or, might they see images of Big Brother, and be repelled by a company snooping on the data they generate as they work? Predictive analytics can enable a customized employment value proposition that maximizes mutual benefit for organizations and their talent; but at what point do predictive analytics become too creepy?

Predict What Employees Will Do Without Freaking Them Out – John Boudreau, Harvard Business Review (5 September, 2014)

I think predictive analytics are creepy, full stop. I’m just okay with it up to the point when some bloke – you know it’d be a man – says anything to me like “Our data analysis predicts that you will soon get restless” and says it with a straight face.

But if figuring out that someone is going to leave means bosses take steps to keep you happy, I’m good with that. The full piece goes on to show that it’s worth these company’s time and investment in analytical software. Unfortunately, it also goes on to creep me out more:

Consider this object lesson from marketing. Pregnancy is an event that changes otherwise stubborn purchasing habits, so retailers want to know about a pregnancy as early as possible. Duhigg’s New York Times story reports that Target marketing analysts built a predictive algorithm to identify pregnant customers based on their purchasing habits and other demographic information. They sent those customers ads for pregnancy related products. What could be wrong with helping pregnant women be aware of products or services they need, as early as possible?

Apparently, women responded negatively if it was obvious that they received pregnancy ads before they revealed their pregnancy. They responded more positively if they received “an ad for a lawn mower next to diapers.” Duhigg reports one executive saying, “as long as a pregnant woman thinks she hasn’t been spied on, she’ll use the coupons…As long as we don’t spook her, it works.” Duhigg also reports that Target company executives said the article contained “inaccurate information,” so the story may exaggerate, but the lesson remains: Effective predictive analytics depends on how real people react, not just on the elegance of the analytics.

Let’s just repeat a sentence there: “Apparently, women responded negatively if it was obvious that they received pregnancy ads before they revealed their pregnancy.” I am torn between saying “Well, duh” or “You think?”

It’s interesting to me that these quotes are from the same Harvard Business Review article but HBR didn’t spot the connection. If a marketing firm can accurately predict when you’re pregnant, so can a personnel department. If given the same data anyway. So you’re at work, you haven’t told anyone you’re pregnant – because you’re never going to tell anyone before the first 12 weeks, are you? – but your boss knows. I don’t like where that’s going.