People analytics, a data driven way to managing workers, is an up-and-coming field that spans everything from human resources to finance to education. The general idea is to use data and analysis to make better decisions on people-related issues.
The Wharton business school at the University of Pennsylvania held a two-day powwow on people analytics and there are a few takeaways worth noting. Here are five things to know about people analytics:
Data can improve communication. Visionary leaders speak in present tense, use simple and clear language and use language that allows the audience to experience the vision, according to Noah Zandan, founder of Quantified Communications, a company that analyzes speech and communication. Zandan crunched the data communication ranging from conference calls to TED talks to keynotes and found those generalities. Quantified Communications' software analyzes your communication based on improving things such as empathy, visual cues and behavioral cutes. one key question posed by Wharton professor Lori Rosenkopf revolved around what happens if everyone starts talking like a visionary. Zandan noted that there may be an issue with actually executing those plans.
Algorithms, behavioral science can improve recruitment. Kate Glazebrook, principal advisor of The Behavioral Insights Team, a joint venture between the U.K. government and private sector, said her firm created an algorithm aimed to sift through inherit bias. The general theme was to get away from making recruiting decisions based on CV strength. To recruit 12 new workers, BIT's first screen was a 700 multiple choice test that whittled the field to 160. Those remaining people were judged based on CV strength by a team and an algorithm that was designed to predict performance. What BIT found was that the algorithm screen called Applied was the most statistically significant predictor of performance. After final interviews, 12 took offers and 60 percent of them wouldn't have made the cut based on the CV sift. Of the five top performers in that group, three of them wouldn't have made the cut under the old model.
Your enterprise social graph is critical data. Microsoft Research as well as a company called TrustSphere talked about tools that extract data from employees based on who they interact with and how long. TrustSphere's software taps into Salesforce, Microsoft Office 365, Google Apps and other corporate tools to create an enterprise social graph. This graph is used for sales (who in the company interacts with a key client the most), risk and compliance and merger integration. That latter use case is interesting. Say a large company merges. And team A and team B say they both have the best relationship with a top client. TrustSphere's software will map what team is legit. Microsoft Research is using mapping to predict employee satisfaction as well as team productivity. In both cases, email logs and anonymous data that rides above actual content can tell an enterprise a lot.
Job description: Data scientist (Tech Pro Research)
Diversity is meeting analytics. Diversity is a theme that surfaced repeatedly at the people analytics conference. One of the big takeaways here revolved around diversity that you can see (age, gender, race) vs. what you can't (background, economic strata, university, view of the world). Analytics is being used to close diversity gaps and frame issues, but what enterprises are after is really having a diverse team that can innovate and collaborate. Diversity of ideas is a key goal. People analytics can help address some diversity issues, but tracking ideas when everyone defines the issues differently is tricky. How do you address quantifiable diversity and traits that can't be measured? "The complexity of diversity makes people hesitant to apply analytics," said Linda Chen, principal at Mercer. In other words, diversity is a messy problem for analytics, machine learning and algorithms. "It's not just gender. It's race. It's religion. It's a whole package. That's a messy issue to tackle," said Ashleigh Rosette, a professor at Duke's Fuqua School of Business.
Analytics can't do everything. Simply put, analytics aren't going to make humans analytical enough, ask the right questions and be creative. Yet, enterprises still may be lulled into thinking analytics are a magic bullet. "We know there's an overconfidence problem," said Cade Massey, a Wharton professor. Massey said the worst case scenario is that enterprises begin to think analytics can do everything.
Tim Urban, the writer and illustrator behind Waitbutwhy, said that a bunch of reports may not reveal much. In fact, people analytics may just be a recipe for followers. "I'm worried that people analytics will favor cooks and not really show what the chef is doing. Chefs may not shine," said Urban. In other words, analytics only make sense if the person behind the data are creative and ask the right questions, said Kathleen Hogan, Microsoft's chief people officer. The other thing that analytics can't do is to get humans to think in statistics and not narratives. Michael Mauboussin of Credit Suisse noted that "normal people don't think in statistics." "The stories swamp base rates and statistics in the real world," he said. "We have communications problems with analytics."
Daniel Altman, an NYU professor and founder of North Yard Analytics, said analytics can predict "goodness of fit," but it isn't done enough. Altman specializes in sports analytics and noted that teams need to quantify overlapping skills between players and how they fit together. "It's about putting the puzzle together," said Altman.
Bottom line: Analytics may make you smarter, but culture, communications and lack of creativity can limit its positive effect.