Startup Humanyze's 'people analytics' wants to transform your workplace

Tracking social activity at work could help companies manage and boost collaboration

The lowly coffee machine may be a common company gathering spot and an essential part of many workers' daily lives, but few would consider it a management tool capable of bringing about organizational change.

That, however, is just the kind of thing that becomes possible when you begin using "people analytics," according to Ben Waber, cofounder, president and CEO of Humanyze.

Waber's Boston-based company has developed a wearable device called the Sociometric Badge that can be used to track people's social activity at work -- where they go within the company, who they talk to, and how they interact. The goal is to help companies understand their employees better, with an eye toward optimizing communication, collaboration and innovation.

In one recent case, an organization created a robotic coffee machine that could act on data collected by Humanyze's badges, Waber said. By linking into Humanyze's application programming interface (API), the coffee machine could "see" when a particular group needed to interact more. In such cases, it could wheel itself into a central position within that department to facilitate that goal.

The result: More group time spent around the coffee machine, more talking and more collaboration.

Or so the thinking goes.

'We didn't know what we were doing'

Humanyze's Sociometric Badges are based on technology from the MIT Media Lab, where Waber and several fellow graduate students set out back in 2006 to develop a better understanding of how people interact.

Recognizing that traditional methods -- surveys and consultants -- had serious limitations, they were looking for alternatives.

"When we were at MIT, we started to realize that people were actually already carrying devices that were measuring an awful lot about what they do," Waber explained.

Cell phones and RFID-equipped company badges, for instance, open the doors to a wealth of information. "If I put RFID readers in the ceiling, I can tell where people are," he said.

Of course, that still doesn't tell you how people are interacting. With that goal in mind, Waber's team began creating a device that added two microphones along with Bluetooth, infrared and an accelerometer as well.

The unit's microphones didn't record what people said. Instead, they tracked things like how long they spoke, how often they interrupted each other and the volume of their speech.

"Imagine you're watching a foreign film and you turn off the subtitles," Waber said. "That's the kinds of things it was picking up on."

In early experiments at MIT, the technology could be used to predict with 85 percent accuracy who would "win" in salary negotiations in controlled test settings. It achieved similar success in tests involving speed dating.

It wasn't long before Peter Gloor, a professor in MIT's business school, caught onto the work being done by the team, and he connected them with real-world enterprises for further testing.

"We didn't know what we were doing," Waber recounted. "We collected data for about a month and decided to look at the simplest thing: who talks to whom."

Their initial goal was to examine those in-company interactions to see if they shed any light on how productive or happy individual employees were. Turned out, that one bit of data alone had enormous predictive power.

"We were doing six times better than all the other data being collected by email, surveys, etc.," Waber said. "The company ended up reorganizing based on our data."

Further use by other companies helped Waber's team refine and validate the device further, and finally, in 2011, they made it official and launched Humanyze. Among Waber's cofounders are Daniel Olguin Olguin, now the company's COO; Tuomas Jaanu, its CTO; Taemie Kim, chief scientist; and MIT Professor Alex "Sandy" Pentland as scientific advisor.

"A lot of us were planning for careers in academia, but I returned from a trip and found I was CEO," Waber joked.

Not that he minded. "Originally we were in this to create a tool to understand how people work," he said. "Now, it's being used to change how companies operate."

Four kinds of sensors

The device has evolved considerably over the years, including becoming much smaller in size -- the original version looked like a tie, but the latest iteration is more like an RFID badge roughly the size of a deck of cards.

It's still powered by four key types of sensors.

First, its Bluetooth sensor scans other devices, so if two people are both wearing badges, it's possible to see who is close to whom as well as their location within the workplace.

The infrared scanner focuses on line-of-sight communication, revealing when two people are facing each other.

The two microphones still don't record what employees say; instead, they perform real-time voice processing to uncover variables like tone of voice and volume.

"We can tell not just if people are in a conversation but also dynamics like interruption, how long each participant speaks when it's their turn, and whether they're speaking differently than they do in other conversations," Waber said.

Finally, the badge's accelerometer measures movement, which indicates how engaged people are in their conversations and how in tune they are. Mirroring behaviors, for example -- such as when one person's forward lean sparks a similar movement in the other -- are a sign that people are in sync.

Using the cloud, Humanyze combines the results of all those sensors with data from other sources, such as electronic communications, objective productivity metrics or key performance indicators -- "whatever the company cares about," Waber said -- and presents the full suite in aggregated form via a management dashboard.

Equipped with such information, management can align internal metrics and resource planning based on key behaviors. It can also adjust a team's structure or environment to support people and their work.

Bank of America tapped Humanyze's technology early on to address productivity and turnover among call center employees. The analysis found a lack of social engagement among teammates, so Humanyze proposed specific changes in team scheduling. Measurements taken several months later revealed sustained improvement, leading to plans for a broader rollout.

Companies can also use the technology to perform a/b tests to help them choose a compensation system, for example, or the best group-chat tool for their team. In such scenarios, one alternative would be rolled out to some people while another would be deployed for others; management would then analyze the data collected by Humanyze's badges to determine which option leads to better behavioral results.

"The head of IT might say, 'I want to move to this new system; it costs more, but I think it will really improve performance,'" Waber explained.

Whereas companies often evaluate their employees on "myopic" metrics such as reducing costs, the technology enables them to focus instead on "how those decisions influence KPIs and how people work," Waber explained.

'We don't record time in the bathroom'

Humanyze has signed up 12 organizations as customers so far. Along the way, it's found that shifting organizational decision-making to this data-focused approach requires some cultural change, Waber said.

"When it comes to the people side of the business, the way things have often happened is that someone reads an article about Google and says, 'Google is a cool company -- we should do what they do,'" Waber explained.

Now, they can use data to guide their decisions.

"It helps people understand the big picture," Waber said. "They can have a much clearer conversation around what's actually at stake."

Of course, none of those benefits are possible if the people asked to wear Humanyze's Sociometric Badges don't cooperate. To allay their concerns, Humanyze takes pains ahead of time to reassure employees that their privacy won't be compromised.

Rollouts take about four weeks, with plenty of time dedicated to that kind of education.

"We first meet with users face-to-face," Waber said. "We give a presentation on what data we collect, and we give them a consent form that shows them the database tables."

It's a legal document between Humanyze and the user, and the majority of employees sign it: The company has had more than 90 percent participation in every rollout it's done so far, he said.

"We don't record what you say, and your boss doesn't get to see your individual data," Waber stresses. "We also don't record time in the bathroom."

Managers can see aggregate data at the team level, but "not what Bob was doing at 2:30 on Thursday," he said. "That's part of our contract with users."

Still, for those who aren't comfortable, there are options. Users can opt out, and they can even get a "placebo" badge so that it looks like they're participating when they really aren't.

'It's like a Fitbit for your career'

For those who do participate, the benefits can be significant, Waber said. In addition to dashboards for management, Humanyze allows each employee to set up their own, personalized feedback reports including metrics such as body language or how much time they spend in conversations, for example.

"It's like a Fitbit for your career," he explained. "When you set up your dashboard, you tell us what you want to achieve."

Someone who wants to be the company's best salesperson, for example, can use the technology to benchmark their own performance against that of the current top performers without ever knowing who those people are.

Alternatively, someone who wants to become a manager can set up a dashboard that uncovers what he or she needs to do in terms of behaviors to achieve that goal.

No matter which department is using Humanyze's Sociometric Badge at any given time, IT plays a central role, Waber said.

"As companies become able to culturally assimilate this kind of approach, IT can go beyond just supporting it and help to supercharge it," Waber said.

Much the way that robotic coffee machine responds in real time as behavioral data streams in, so other IT tools could adapt the way they operate based on the data, he suggested.

Programmable lighting and flexible office layouts, for example, could be automatically configured to "nudge" employees in the organization to behave differently, for example. Email could be delivered more slowly when an individual's workload is high, and certain emails could be emphasized when communication with a particular coworker is overdue.

In addition to offering its Sociometric Badges and consulting services to customers, Humanyze also provides its technology on a not-for-profit basis to some 100 research groups in the interests of exploring its potential further.

One group, a team from Arizona State University, produced some particularly intriguing results. The team's focus was creativity -- in particular, whether it could be measured behaviorally.

"For a long time, researchers have studied creativity using a variety of tests," Waber explained. Surveys and interviews, for example, are among the tools typically included in that time-honored but time-consuming toolkit.

This time, the ASU group used Humanyze's Sociometric Badges along with the standard tools to assess creativity at numerous R&D labs across the United States.

"The question was, if we combine all those traditional ratings into a number -- representing our best possible guess at how creative people are on a given day -- what are the specific behaviors that predict that rating?" Waber said.

What they found was that two key behaviors measured by Humanyze's badges could predict that creativity metric with more than 90 percent accuracy: How physically active the person is at work, and how diverse a group of people they talk to on the job.

"Companies today have all these initiatives to promote innovation," he said. "But we can say pretty confidently that if they don't affect either of those behaviors, they're probably not going to have an effect."

Join the Computerworld New Zealand newsletter!

Error: Please check your email address.

Tags emerging technologycollaboration

More about Arizona State UniversityBank of AmericaGoogleMITMIT Media Lab

Show Comments