10. The Future of Organizations: How People Analytics Will Transform Work

The modern organization is truly amazing. Thousands of years ago, humanity could do little more than produce a few simple sailing vessels, rudimentary weapons, and singular artifacts. Some of the organizations that produced these items, however, consisted of hundreds or even thousands of individuals.

Eventually, humanity gave birth to the formal field of management, which tries to guide the development of products and services under a more scientific framework. Companies have continually accelerated the process and scale of development. Emerging from that vast tidal wave of history and evolution and accident, the wave that finally touches the shore is the modern organization.

Over the last few decades, information technology has had a major impact on organizational design. By changing how people communicate, IT has enabled companies to consider new ways to collaborate at work. Before the Internet, collaborating on a document with a team in another city was unthinkable. Because experts can be anywhere in the world, however, collaboration through e-mail became a desirable option. The rise of mobile telephony and real-time video chat enabled companies to open extremely small branches all over the world, further relying on local expertise and rapid, rich communication between different teams.

Certainly, further advances in IT will occur, and they will have similarly profound impacts on the way people organize, but none of these tools are reflective. That is, they don’t feed anything directly back into the organization.

Think of a company as a water system. The formal and informal parts of the organization are like the pipes. Some pipes come from informal processes such as bumping into people around the coffee machine, whereas other pipes represent formal reporting relationships. If you can structure the pipes correctly, the right amount of water will end up getting to the right destination. Otherwise, you’ll burst a faucet in a house in Kentucky because you routed all the water for the state of New York to a small outlet, while other parts of the country experience widespread drought because you siphoned off too much water.

IT is like the pump. The better it performs, the faster it can push water around the pipes, and the quicker you can adjust to any change that emerges at the ends of the system. Better pumps might also lead you to build new pipes and remove old ones to take advantage of the additional pressure and speed that the new pump affords.

However, the pump will never tell you whether you laid your pipes correctly in the first place. Companies today have a very good understanding of the pump and what comes out at the end of the different pipes, but they have no idea what’s going on inside of those pipes. This is where the next frontier of management lies: in using sensing to change organizations.

What this book has covered is how sensing technology, and big data about organizations in general, can have massive effects on the way companies are organized. From changing the org chart to changing coffee areas, no aspect of organizations will be untouched by the widespread application of this data.

Because this technology is so new, the scope of the studies examined in this book has been limited to single companies. I hope that by expounding on some of the general lessons from these projects, I’ve been able to convince you of the transformative power of data, particularly communication data. This work will expand in the future—already dozens of research groups around the world are applying this sensing methodology in their work.

Imagine, however, that in the future Sociometric Badges and the other methods you have seen used in this book aren’t just limited to a few companies and some academic projects. What if the Sociometric Badge became everyone’s ID card? What if instead of thousands of people wearing badges, millions of people were wearing badges continuously for decades? What things could we learn about how to better manage companies? What new opportunities would arise?

Ease of use is relatively simple to tackle, because electronics are always getting smaller, cheaper, and faster. Most company ID badges already have an RFID chip inside, which is essentially a very cheap sensor. Even the current version of the Sociometric Badge is getting close to these ID cards in size and weight, with the battery remaining the biggest bottleneck to further miniaturization.

Using sensors and calculating voice features takes power. This is why your cell phone runs out of juice more quickly when you turn on Bluetooth and GPS. When you transmit a signal over the air, you’re effectively blasting light waves in all directions. This is what the badge does to help detect conversations. On top of that, voice features are constantly computed when someone is speaking, requiring signal processing algorithms to run on the badge’s processor. When someone isn’t speaking, a lot of the processor’s parts can be turned off, further saving power. The point is that all of this activity takes a lot of juice. The current version of the badges does pretty well: The battery lasts for about one work week without needing to be recharged.

Recent innovations in battery technology, from inductive charging to over-the-air power transfer, mean that in a few years the badges won’t even have to be directly recharged. They can be charged simply by placing them on the table for a few minutes or even wirelessly beaming electricity from an in-office power station. With some clever engineering, the power consumption of these sensors could be reduced even further, until soon people are using these sensors as their regular ID badges.

The privacy portion of the equation, however, gets more complicated. When sensing technology becomes commonplace, the Sociometric Badge won’t be the only option. Other companies could make similar devices with similar capabilities, but instead of adhering to a participant-centered version of privacy as discussed in Chapter 1, some of them would likely have a different approach. If the badges are now your ID, after all, will people still have to opt-in to data collection, or would companies turn it on by default?

These questions are difficult to answer and indicate that strict legal mechanisms must be in place to ensure that companies don’t have access to individual data. Although pockets in the legal community that are interested in establishing such standards exist, such as the Berkman Center for Internet and Society at Harvard, this movement is only slowly gaining traction. Without these protections, sensing technology could create a poisonous work environment, one in which people are constantly worried about being spied on and monitored down to the tiniest movement.

This type of environment completely defeats the purpose of the badges and organizational sensing in general. The badges are designed to help make people happier and more productive, things that all companies should strive for. There’s just not a good business case for checking what Bob was doing at 2:30 p.m. on Tuesday. Not only is exposing that information a huge violation of privacy, it also wastes the company’s time. The things that all companies should care about are

• What makes people in this company productive and happy?

• How can the company change to make people happier and more productive?

Notice there is nothing about individuals, because that data is too specific to affect broad change across an organization.

The issue of privacy deserves a book in and of itself, so not a lot of time can be spent on it here. Luckily, some great work by the World Economic Forum, Personal Data: The Emergence of a New Asset Class,¹ outlines the challenges in this space and describes in detail the framework that needs to be in place for innovation to continue. It’s heartening that governments and businesses at such high levels are paying attention to this issue and agree on the basic “new deal on data” principles.

You should feel more confident that not only will technology such as the Sociometric Badges become widespread, but it will also be adopted in a privacy-preserving fashion. So what will happen after millions of devices are scattered across the globe? Let’s take a tour of what this would look like, from integrating new employees to the fundamental building blocks of management.

Having hundreds of thousands of employees at the same company wear badges for years on end can multiply these benefits. Every facet of the organization can be analyzed and modified to be more effective. Although this book discusses many of those areas, these changes could be made at an even larger scale.

Let’s now begin where all companies do: hiring employees. On-boarding new hires is one of the most difficult things that companies have to do, with the common statistic being that bringing on a new employee costs about 25% of that employee’s salary. Using the badges, however, could affect this process at a basic level.

Consider training employees at a retail store. They have to learn the culture of the organization, how to restock the shelves, how to work the cash register, and how to interact with customers. Sensors could be injected into every part of this equation. Reading RFID tags on clothing could help employees track their restocking speed (this technology is already used at many warehousing facilities), and cash registers could time transactions to give feedback on speed and errors.

The big payoff, however, would be in cultural integration and customer interaction. On the culture side, employees could look at how integrated they are in the social fabric of the organization. This would show how cohesive their group was, and the people analytics system would automatically suggest ways that they could change their interaction patterns. For example, if they’re eating lunch alone, then going out to lunch with some of their colleagues might help build cohesiveness. This data could also find its way into shift scheduling, so that the right people are paired up with each other when they’re taking inventory, manning the cash registers, or stocking shelves.

Improving interaction with customers would have a direct impact on their paycheck, because employees in many stores are partially paid on commission. Learning how the most effective salespeople interact with customers, in terms of tone of voice, volume, and speaking speed, would all be extremely helpful to new and veteran employees. Employees could work on voice pitch in real time by talking to the computer to gauge their speaking style, or look at feedback reports afterward to check their progress.

At its core, showing employee progress and disseminating best practices automatically would be a huge contribution of the people analytics system. Today these practices occur through an arduous process of trickle-up reporting. If an employee at one store figures out an effective way to interact with a customer, it has to first be noticed by his boss, then reported up to her boss, and up and up and up until it eventually finds its way into the employee training program. After this practice is codified, however, feedback on how an employee matches up to that ideal is subjective or even non-existent. Most companies act as if employees are veterans after going through the training program, but of course there is normally a learning curve that continues for years—not to mention the fact that best practices can frequently change. If a store starts selling a new kind of product, it can take a long time for new best practices to make their way back up the hierarchy.

Beyond the individual level, communication data allows companies to look at how people work together and how to help them do that effectively. Even before putting a team together, however, companies could simulate how well they can expect employees to work together and what challenges are likely to emerge. As discussed in Chapter 3, communication dynamics are the lifeblood for team success. Some projects need a group with diverse connections to gather new information quickly, whereas other projects need tight-knit connections to execute tasks effectively. Companies could test out different team configurations beforehand to maximize these different traits and be more confident of their success. Data from sensors and digital communication records would help determine what tasks different teams would be good at, creating a sort of “team fingerprint” that employers could match with certain task types.

Another important benefit of teams is that they create long-term social capital. In other words, if you work on a team with someone, you develop a deep relationship with that person. Down the line you can call on that person if you’re working on another project that requires her expertise. Passing along information that’s relevant to you is also easy for her, because she knows what you’re interested in. Constructing a team fingerprint is then not just about how a team would perform on a specific project, but how it would benefit the company down the line. Using sensor data to drive these predictions would make the team fingerprint a major driver of organizational success.

After a team is formed, giving members a sense for how their dynamics are changing over time is also important, particularly as it relates to different phases of a project. These dynamics could even be shown at the meeting level, using a real-time feedback system to continuously tune participation dynamics over long periods of time. People who tend not to participate in meetings could be visually encouraged to jump in by seeing their participation bar flash, while people who were overly dominating discussions could also be flagged.

In fact, my colleague Taemie Kim at MIT developed an early version of this system called the Meeting Mediator. Using data from the Sociometric Badge, this system showed participation levels and communication patterns for each meeting participant, encouraging dominant individuals to pull back and less talkative people to speak up. In a number of studies, from trust games to brainstorming tasks to decision-making scenarios, Taemie was able to show that teams who used this system trusted each other more and cooperated more effectively. Imagine the kind of influence this system would have if it wasn’t just used as a one-off, but became part of a team’s culture.

At a more general level of communication, the people analytics system could show a team’s balance between exploration (communicating with people in diverse groups) and execution (having a tightly coordinated collaboration pattern). Depending on the needs of the team at any given moment, this data could deliver suggestions for how to shift this balance as well as make slight modifications to the work environment.

Teams that are actively trying to gather new information, for example, might be encouraged to ask for introductions from certain colleagues. The system could even start sending invitations for social events that other groups are holding if the data says this would be beneficial for all involved. The point is not to create awkward social moments where strangers show up uninvited, but to make people aware of opportunities to meet their colleagues in new ways.

With this proactive and intuitive organization, the formal org chart will rapidly fade in importance. Although in the past the org chart was useful to help channel communication and collaboration, with widespread adoption of sensing technology and communication data mining, the focus will change to creating an environment that nurtures the connections between people. To make this idea more concrete, imagine a company wants to open a new line of business. Previously this would involve creating an org chart, spending months figuring out who would fill out that organization, and developing a strategic plan to guide the new division.

That strategic piece will likely remain, but now imagine that this sensing-based organizational dashboard is deployed across the company. The people analytics system observed many new divisions open up over time, and has seen what factors lead to success from a collaborative perspective. After the basic team functions for the project are chosen, the system would know with a high level of accuracy how the different teams should be connected, how the teams should collaborate internally over the course of the project, and what factors allow those effective communication patterns to flourish. More importantly, this could also change over time. Rather than having a set org chart, collaboration patterns could rapidly change depending on issues that arise during the project.

You’ve already learned about some ways to affect communication patterns, but let’s figure out how these would all connect to the people analytics system. One lever at our disposal is the physical office layout. As goals change, the system will know the office layout, in terms of social areas, furniture, and desk location, that will best achieve those goals. Coffee area and café locations are important drivers of collaboration. Desk location and even the type of cubicle people are in, as you saw in Chapter 4, often influence who people talk to more than any formal requirements. Clearly, ripping out old cubicles and putting in new ones every week, or even every few months, is impractical. Coffee machines, however, are easily moved. Desks can also be changed every few months without much interruption, especially if they are on wheels.

Imagine that on Monday the system sends out an e-mail to a team member suggesting a new location for the coffee machine. Walking through the office in the morning and seeing the coffee machine cart migrating to its new home could become a normal sight. Imagining a “moving day” where people across the office spend a few hours swapping seats in a semi-chaotic game of musical chairs is also not difficult. This practice is actually not so different from what happens in offices today. Offices are frequently reorganized based on changing constraints. The difference is that with the people analytics system, data drives those changes, making internal migration a part of the corporate culture.

In offices with open seating, we don’t even have to encourage internal migration. People already choose their desks every day, although normally their choices are based on gut instinct. The people analytics system could sit on top of this process and send everyone seating suggestions in the morning to optimize collaboration patterns. People could still choose whatever seat they wanted, but now they could use hard data to make their decisions.

The sensing technology can also be incorporated into the environment itself. As an experimental project at MIT, Alex Speltz and I developed what we called an “augmented cubicle.” The cubicle itself had the same dimensions as a standard cubicle wall, but instead of being lined with beige fabric it was composed of a window shade sandwiched between two panes of plexiglass. The window shade was connected via a wire to a small motor in the base of the cubicle wall, which was in turn connected to a small computer that we could communicate with over Wi-Fi.

The motivation for building this wall was to make people more or less visible to others based on their social context. So if it seemed like one group in the office needed to talk more with another group based on project dependencies or long-term social capital factors, then at night a program would signal all the cubicle walls between the two groups to raise their blinds. When everyone came in the next day, you could walk by someone’s desk and easily see what they were doing and seamlessly start chatting with them. Of course, a manual option would be available to move the shade up or down when workers didn’t want to be disturbed, but in general most people stick with defaults.

The shades wouldn’t change more than once or twice a week, and they would only move at night. This practice would preserve a natural office environment while taking advantage of the rapid changes possible via a people analytics system and a bunch of motors. This concept might seem like one that’s far down the line, but the idea that an office could be reactive to its inhabitants in a non-intrusive way is a powerful one. Most likely this approach will start working its way into all of our lives in the not-too-distant future.

Another lever to consider is communication tools. This book has examined some of the ways people can communicate today, but new channels are constantly added to our repertoire. Internal Twitter clients such as Yammer have started to gain in popularity, and consumer tools such as Google Hangouts are quickly making their way into the corporate world as well. Add these onto face-to-face, phone, IM, chatrooms (which are still refusing to go away), discussion boards, teleconferences, and wikis, and the number of communication channels starts to look overwhelming.

This extreme power of choice is not necessarily a good thing. If you work at a company where different groups use all of these different tools, then you might find it impossible to talk to each other. The marketing group might be composed of heavy IM users, for instance, while the finance people could be wiki devotees. Instead of making communication easier, the proliferation of tools can have the opposite effect.

This does not mean that more choice is always bad, or even the fact that certain groups use specialized communication tools is bad. For example, allowing software developers to communicate within their development environment makes them much more effective. It helps them coordinate with their colleagues as well as automatically recognize problem points in software projects. However, a data-driven approach is needed to help companies figure out what tools to use and how they best support different kinds of interaction. By plugging a people analytics system into these tools, companies could roll out new tools across the organization according to communication needs or move people to more effective media by merging other tools.

A people analytics system could also be used to modify how the tools themselves actually work. In a project that I did while I was working in Japan, your relationship to the sender would actually change how e-mail was displayed. For example, if increasing the diversity of your connections would be helpful for you, then e-mails from people in other social groups would increase in size, literally standing out from all the other mail in your inbox. This slight visual nudge would be a reminder that reaching out to those other people might be a good idea.

This feature doesn’t just apply to e-mail; it could also be used to display people differently in chat windows and discussion boards. Teleconferences could be similarly altered by encouraging people who aren’t speaking to participate by amplifying their handset volume or ratcheting down the volume of others. These changes wouldn’t force a change in behavior by itself, but would help push individuals and teams to slightly alter their communication in positive ways.

The people analytics system could also send messages over these communication channels to shape the social structure of the organization. Some general parameters might exist for how connected the company should be—that is, how many hops in the social network it takes to get from one team to another. What if you could use a people analytics system to create those ties?

The basic concept here is fairly straightforward: Let’s say I want to connect two people, and they have a friend in common. The best way to connect them would be to contact their common friend and say: “It seems like it would be helpful for your friends if you introduced them.”

This method has a couple benefits over directly connecting these two people:

• The system would have a hard time figuring out whether the two friends would really hit it off (a lot of variables go into that); however, the common friend knows both of them, and probably has a good idea about their compatibility.

• To the people being introduced, having a common friend make the introductions seems like a natural social process. Friends get introduced all the time. There’s just a little bit of data getting injected into the process.

This whole concept is called Augmented Social Reality. Augmented Reality refers to a field of wearable computing where programs overlay information onto video images from your phone or through special glasses. This practice has become popularized through augmented reality apps such as Layar, which you can use to see nearby tweets or the Yelp rating of restaurants, all layered on top of the live video-feed from your phone’s camera. Augmented social reality, however, is about using sensor data to turn everyone into a social connector by layering social context on top of our everyday interactions.

Some people are just naturally good connectors, but it’s an extremely difficult skill to master. By creating an augmented social reality, a layer on top of our work lives that allows us to see who we should be introducing and how to interact with people, everything flows a lot more smoothly. This framework could be used to slowly stitch together far-flung parts of the company, socially knitting them together. By going from introduction to introduction, a system could introduce one team to another that’s just a bit closer to a more distant team, then making another introduction that gets them a bit closer, and so on until the teams are finally directly introduced. This process is not going to happen overnight, but making it a part of the organizational culture, a normal tool that people rely upon, could profoundly shape the network into a much more effective whole.

A people analytics system would likely find its way into other organizational processes as well. Rather than giving formal groups bonuses based on performance metrics, people from other teams who informally participated in a project could be rewarded. HR evaluations, which typically use surveys and qualitative reports from managers, would naturally benefit from an infusion of behavioral data. Organizational strategy choices could be weighted by information similarity from different team members, so that people who have a more diverse social network would have a greater voice than people who were just participating in an echo chamber. The list of applications goes on and on. In any case, one thing is clear: People analytics will radically transform the way companies do business. But for users, for workers, it will become familiar and commonplace.

What about the variation between firms? IBM, for example, could try to change the way it behaves based on how Google organizes its people. Today, this type of change happens from people reading news articles, case studies, and books like this one. These are all great ways to disseminate best practices, but making the change is an incredibly slow process. It takes years before people from outside companies discover new management styles in other firms, and it can take another few years before these styles seep into the popular consciousness.

One approach to help speed up this process is to create organizational benchmarks that use data from companies all over the world. One of the most popular benchmarks in this vein is Gallup’s Engagement Survey, which consists of 12 questions that they have asked to millions of people in thousands of companies all over the world. Companies participate because they want to see how they stack up against their competitors, and how they can learn from those competitors to improve on their weaknesses and capitalize on their strengths.

Now imagine this benchmarking method applied to the kind of sensor data discussed in this book. Instead of just keeping best practices within companies, lessons learned in one company can be applied to a completely different company, all in the blink of an eye. For example, suppose you’ve observed a hundred drug development projects at pharmaceutical companies, each with varying levels of success. You could relate that success to changes in collaboration patterns, showing that shaping a team’s interactions over the course of the project can make them successful. Now whenever another pharmaceutical company starts up a project, managers know exactly what kinds of interactions to support without needing any additional input from the company.

All of this information sharing is incredibly exciting, because quickly companies in India would learn from companies in Brazil that would learn from companies in the U.S. Organizations in totally different industries with vastly different business models would be able to rapidly exchange best practices, taking advantage of similarities between their businesses inferred from hard data. The people analytics system becomes a global learning network, where the management “experts” no longer tell companies what to do, but instead the combined power of data drives how companies are organized.

This system might seem a bit scary to some managers, because it applies the same tools to management that one typically associates with automated hardware and software fixes. Management has always been a domain where soft skills are needed to make it work. This people analytics system, however, democratizes the management process.

This democratization is crucial for an area that hasn’t been discussed much: small business. Companies with only a dozen or fewer people typically have very few tools at their disposal to help them manage their company. This problem is compounded by the fact that most people who open a small business have little management experience. These small businesses aren’t just a sideshow in the economy, either. About half of non-farm GDP in the U.S. is generated by small businesses.

The people analytics system would essentially be “management in a box” for small businesses, enabling them to apply to their company the lessons learned from other organizations. With only a few sensors and some basic programs, people in small businesses could get automated help setting up their management structure and generating effective collaboration patterns. They could even receive feedback on their progress. Not only that, but with enough small businesses using such a system, these fledgling companies could get automated suggestions on org structure, compensation systems, and so on.

Take, for example, a bunch of friends going into business together to open an upscale pizza restaurant. They get the lease, purchase all of their cooking equipment, and finally buy their “People Analytics Starter Kit.” They fire up the system, input the type of business and what each employee does, and then hand out badges to everyone. Immediately the system suggests different ways they could manage their business, from compensation systems to org charts. However, the real payoff comes when they start collecting data.

Every day the system provides automated feedback to the restaurant. It seems like the waiters have really engaging conversations with customers but tend to interact with customers less frequently than the highest performers at other establishments. It also seems like the head chef isn’t checking in enough with the line cooks, which means more order errors and slower cooking times. Even in a restaurant, the chain of communication is essential to make sure that the order is going from the customer to the waiter to the head chef to the line cooks in a timely fashion. Measuring that response time and giving feedback on how to improve it could be the difference between seating 80 people for dinner and seating 100. After seeing this feedback, the waiters check on patrons every seven minutes instead of every five, and the head chef now chats with other cooks every few minutes to see what they’re up to. Customers are happier, the restaurant runs better, and the results are fed back into the global system so that next new restaurant will work even better.

A people analytics system would create a learning community of organizations the world over, not through reading articles or books, but by exchanging data. This data exchange would extend from the Walmarts of the world to the mom-and-pop stores on the corner. From companies that install augmented cubicles in hundreds of offices to businesses that just want some simple pointers, this feedback will change what it means to manage an organization. People analytics will be here to stay.