Blog

Persistent Insights

March 10, 2016
Enlighted Inc.

Clifton Lemon is a well-known lighting and building guru with a keen eye for the future. He has been contributing a series of articles to our blog on the future of lighting and his thoughts about some of the implications of IoT. This is the seventh of ten articles.

Embedded Behavioral Analytics Redefine Evidence Based Design

A promising and surprising trend uncovered by embedded sensor and analytics systems like Enlighted’s Space application is that previously unquestioned assumptions about how people use buildings often begin to crumble in the face of granular realtime data and the insights it provides. Space is only one of many current types of applications impacting the design and planning of facilities- there are dozens more potential ones on the near horizon. And the impact of these systems extends beyond improving the end results of design – it can dramatically redefine the most important part of design, the front end research.

Industrial behavioral research as we know it today began with the work of Frederick Taylor, whose time and motion studies were among the first “scientific” methods aimed at counteracting the unintended consequences of ramped up industrial production made possible by electrification. One result of the standardization and rapidly increasing mechanization of industrial work is that people whose jobs used to be diverse and interactive suddenly found themselves with narrow, highly specialized jobs, standing or sitting in front of deafening single function machines, pulling a single lever for ten hours a day as identical widgets rolled by on the assembly line. The unforeseen social consequences of mechanization included devastatingly high labor turnover, drops in productivity, and low morale, and became dealbreakers for the economy. Since behavioral science itself was fairly new and unproven, it was natural that corporations assumed that what worked for machines – rational, cause-and-effect, physics – would work for people too. This viewpoint essentially turned people into machines and did not go over well, to say the least. One result of the violent reaction of the workforce was what we now recognize as corporate welfare, which ultimately became our current health care system, as corporations made more and more concessions to workers in an effort to lower the turnover rate.

Today behavioral and cognitive science has made much progress, but has not yet really begun to see its potential to address crucial problems we now face in the workplace. What we call “building science” is applied to more easily quantifiable physical things – the energy use, materials, systems, and configurations of the buildings themselves, not the people who live, work, and play in them. And few corporations can afford more rigorous research like that conducted by William Whyte in the 1970s, especially for indoor environments, where it’s difficult to unobtrusively observe how people work and socialize in the office environment. Who wants to work with someone standing around all day with a clicker counting people and making notes on the frequency of their trips to the water cooler? Fortunately most of the repetitive, quantitative nature of this kind of data gathering has been highly automated and has become much more accurate, efficient, and affordable. The problems of attracting and motivating workers and stemming turnover are similar to those of a century ago, but facilities planners still don’t have the data and deep understanding they need to tackle productivity. Millenials evidently want bean bag chairs, cuddle rooms, sushi bars, espresso machines, foosball tables, and a balance between social and private space, but how much, where, and most importantly, why?

Behavioral research is a critical part of design and planning but is not always particularly valued in our business culture today. In the past, research meant squadrons of people in white lab coats with clipboards, asking questions and making observations. Research projects were slow and expensive and results were often misunderstood, simply ignored, or worse, based on bad science in the first place. This is one of the reasons why there’s not much in the way of useful behavioral research in the built environment. Also, what’s available is focused almost exclusively on energy efficiency.

One problem with built environment research is an over-reliance on techniques like post-occupancy building surveys, which are typically done with the crudest tools available and characteristically fail to deliver rich insights to crucial behavioral questions. According to this article, such surveys suffer from a range of design problems, including selection bias, small sample sizes, and low accuracy and used alone, which is often the case, are simply inadequate in drawing conclusions about building performance. A key reason for this is that they rely on self reporting and conscious choices rather than independently observed, unconscious behavior. Because of cost and time constraints, these surveys are typically done over the web and less often in interview situations where unskilled interviewers routinely subconsciously influence results.

What to do with all the new smart analytic capability that’s so easy to implement? Let me count some of the ways. For starters, I think we need to forget about the typical way we practice “research” in the built environment today, a proposition that with many risks and unclear rewards. It involves spending a lot of time and money on experts, trying to prove one thing and more often than not accidentally stumbling on something else that’s much more useful. The combination of powerful data gathering systems and analytics can transform the process of learning to make better buildings and free us to focus on the real problems, which are behavioral, not technical.

What I’m going to christen here as the fabulous new Paradigm of Persistent Insights means that with a rich sensor network enabled with a wide range of input devices, crucial behavioral, emotional, and experiential data can be automatically aggregated and analyzed for whatever investigations are important – we can focus on the insights we need rather than the mechanism of collecting the data that allows us to uncover them. In fact, within certain parameters, built spaces can be seen almost as permanent experiments. In many cases there’s no need to devote resources to replicating a retail, office, industrial or institutional environment just so that you can study it closely – you can change things on the fly and observe many direct and interactive effects, simultaneously if desired.

The first obvious example is with lighting, which is visible, increasingly digital, easily controlled, and of course the backbone system for many sensors. We’re constantly increasing our understanding of how lighting impacts behavior like productivity, learning, buying behaviors, and health outcomes. Any retail space equipped with dimmable and color tunable lighting and a sensor based analytic system can get immediate feedback on the effects of changing lighting on foot traffic and sales – this is directly observable data. This applies to offices wanting to test the effects of lighting on productivity as well, of hospitals needing data to document the effect of lighting on recovery times, patient complaints, of fatigue in nurses, for instance.

Or consider the movement of people through a building, which is now highly mappable with Space. With the new flexible open plan (on not) systems, workplaces are highly configurable. What happens to human circulation patterns, social interaction and productivity under different space configurations and adjacencies of amenities? Then consider the interactive effects of lighting, thermal comfort, and circulation. The possibilities are endless, especially when captured data on existing spaces is used to inform the design of new ones. Certain patterns inevitably emerge that many designers recognize intuitively but can’t always articulate with the benefit of quantifiable data.

What will provide some boundary lines of course are limits on companies’ ability and willingness to tinker with their workers like so many guinea pigs and to risk downtime as “experiments” are configured and complex, intersecting insight parameters determined. But with analytic system in place and relatively feasible to implement, what we can now focus on is the important questions that drive business, like employee health and well being, customer experience.

Even if you only want to focus on energy in buildings, as most people do, and don’t want to go in to the murky, scary world of behavior emotions, and experience, you really can’t avoid it. We have learned a great deal about building energy use and how to build much more efficient buildings, but we’re not doing it at the scale we think we need to, and our failure to do so is not based on a lack of technical solutions or advanced materials or software, it’s because we don’t understand our own behavior very well. Why do architects persist in making glass buildings and glorifying this aesthetic in the face of overwhelming evidence of their dismal thermal efficiency? Why do people not use window shades when doing so would save large amounts of energy? Why do people buy more vegetables under certain types of lighting conditions than others? When people claim to be “mostly happy” with their work environment but consistently leave a company after an average of six months, how much of a factor is the workplace, and how do we know? These are the kinds of questions we can begin to answer with a better approach to generating evidence based insights. With embedded analytic systems, companies can begin to answer them with real data and make better design decisions.