This is an exciting time in the history of office buildings. The confluence of three long-term trends—the rise of co-working, the adoption of building technology, and the ubiquity of smartphones—is transforming workplaces. These trends are driving a kind of consumerization, in which users can customize their experience to a significant degree or choose a different space as their needs evolve. While this flexibility is a virtue in itself, there is a second benefit less often discussed: the generation of data through use.
Now that internet-of-things (IoT) technology is being baked into buildings from the beginning of the design process, the relationship between IT and real estate is set to become a fulcrum of business success. In some parts of the economy, this transformation has already been happening for some time; witness the many ways WeWork is turning data about occupants into a core business asset.
In some ways, the period we are now entering is reminiscent of the early days of the web. While the value of internet user data has now been fairly well established, it was not always so. Google grew to become one of the most valuable companies in the world by aggregating and analyzing what many users probably thought was worthless information. This phenomenon essentially repeated itself when Facebook and other social media entered the scene.
Web companies weren’t the original pioneers in seeing the value of user data, of course. That trophy goes to retailers. Just about every chain of supermarkets, pharmacies, and big-box stores has a loyalty program these days; retailers like Target have spent decades cultivating the capability to gather data on their customers. Yet, until quite recently, even those efforts were limited by the tools that were available, which were mostly at the point of sale. For instance, retailers could analyze what customers bought and how often, but it was much more difficult to collect information on what they looked at without buying, or how long they took to decide between two competing products. This is likely to change soon, thanks to innovations that will have implications for all work environments.
A few years ago, Bluetooth-Low-Energy technology (aka iBeacon or Google’s Eddystone) was hyped as a revolutionary technology for retail. It offered a way to collect customer data and deliver content to smartphones seamlessly, based on proximity to any number of cheap digital beacons. While the initial rosy predictions have not really come to fruition for several reasons (one, it started to annoy shoppers), the technology still exists. Its potential in work environments is only beginning to be explored. We are also just a few years away from ubiquitous sensor networks powered by 5G wireless, which could profoundly shape both urban and workplace design.
Workplaces are among the richest data sources available. Many people spend more than eight hours a day in a single building, completing a wide variety of tasks. They use many tools that are data-enabled, or easily could be: secure access points, destination elevators, even smart appliances. When an occupant uses their phone to book a conference room, adjust the temperature, admit a visitor, or change the music in the cafe, they leave behind a trail of data. Properly analyzed, the data can reveal much about the workplace and how people use it, guiding real estate, design, and facilities teams toward better decisions.
Change management is one key area that stands to benefit. When asked, people don’t always tell you what they really think or act the way you might imagine. The discrepancy between stated preferences and those revealed through observation may be large indeed, and this is one area where improved tools to collect and analyze information about occupants can be illuminating for occupants, designers, and managers. The quantified self movement (Fitbit, et al.) shows that making data user-friendly can make people more aware of their own behavior and reveal areas of possible improvement.
Communicating effectively about future or past changes to environments and policies requires good data about what people want and how they are reacting to change throughout every stage of the process. Analyses that make use of smart office technology can transparently collect more information and do so more accurately. When changes in floor plans, daylight access, or other aspects of design are made, the behavior of occupants can be examined systematically. Policy changes can be evaluated based on their demonstrated effects. Longitudinal data collected over the course of months may show the impact on the way people interact or use their work spaces.
In addition to obvious privacy concerns, there are some additional pitfalls that await as workplaces begin to employ data science methods. An interesting property of so-called “big data” is that it can strip away context and nuance. In fact, that’s part of the point—to standardize properties of information so that a computer can crunch the numbers. A famous example of this was highlighted by “the Napoleon Dynamite problem.” When Netflix offered a $1 million prize to the team that could create a better recommendation system, teams were reportedly bedeviled by the cult classic and other quirky films that inspired strong reactions (positive or negative) from viewers. Outliers and edge cases become problematic without context.
There’s also the thorny problem of bias—every algorithm is susceptible. For example, Amazon had to abandon a prototype AI hiring system because it favored male candidates, even though it was built to ignore gender entirely. The humans designing the system also bring their own biases to the table, both real and perceived; employee concern led to the recent dissolution of Google’s AI ethics council over a controversial member. Incorrect assumptions about the meaning of information, decisions about which factors are most important, and failing to consider social factors can lead even well-intentioned efforts to fail.
Supposing for a moment that one can design a system that minimizes these risks, all the usual problems still apply in workplace interventions. Even with the best data, people are prone to:
- Analysis paralysis: They get the data, but don’t know what to do. A company may collect great information about every aspect of the workplace experience. But as the dataset grows, it’s easy to get bogged down in the analysis phase.
- Tone-deaf decisions: They do what they data say, regardless of consequences. An organization may look at badge swipe data and determine that occupants spend very little time in their own cubicles—maybe those people don’t need them! But that doesn’t mean they will feel good about losing that space. Some may think they are being punished with eviction, and even just the suggestion that this could happen might lead them to stay at their desk more when it’s actually not the best way for them to do their jobs.
- They mistakenly address a symptom rather than the cause. Data collected by a room-booking app may show that a certain conference room is not used very often, leading leadership to convert it for another use. What the data does not reveal is that the room is actually just too warm or noisy, or lacks a critical piece of technology.
The only way to find out what is really going on—and what people actually want—is to also look at the “little data.” Part of the design process must involve talking to people and watching what they do. Against the backdrop of the hard math from data science, this may seem like a touchy-feely suggestion. It is not. Social sciences such as anthropology have developed rigorous methods for observational study, survey design and analysis, and interviews that can provide the crucial context that’s missing from all that math—or, more accurately, that the data science solution just wasn’t designed to capture.
In the current climate, anyone who isn’t thinking about data privacy is not paying attention. The regular drumbeat of news stories about leaked consumer information and misused social-media data raises valid questions about public trust. This goes double in a work environment, which is also bound by employment law and cultural expectations—indeed, some are already calling building data use “creepy.” Would-be adopters will have to navigate these murky waters.
It’s safe to assume that co-working operators are already making use of occupant data to improve their offerings and find out more about their customers. More traditional employers are likely to follow suit in the next few years. In the near future, when you arrive for a scheduled meeting, the video conference system will already be set up and ready to go. When you enter a room with a colleague, the lighting will be adjusted to suit the context—a casual conversation or a contract negotiation. When you get up from your desk for a coffee break, the coffee machine will know you like an iced caramel macchiato.
This improvement in service is not free. Smart buildings will soon offer the same customized, exceptional experiences we’ve come to expect from the likes of Uber or Netflix, but the tradeoff is that we share our data. Just as people routinely share their location with a stranger when they call an Uber, denizens of the office will probably choose a decrease in expected privacy in exchange for a better experience. What they get in return is not just the service received, but the promise of better designed buildings in the future, and transforming how buildings work for the next generation.