The fundamental differences between manual counting and occupancy sensors and which is best for your needs.
How can we use our office space more efficiently? This is one of the key questions asked in our industry and one of the hardest to answer. The thin balance between providing sufficient space, to enabling productivity and growth, and wasting real estate has contributed to the creation of more sophisticated ways to help companies measuring and improving their space utilization.
Despite the growing number of companies and devices in the market offering space utilization studies, they all fit in one of these two categories of data collection: manual counting or occupancy sensors. The information you will get will lead you to a decision that will potentially change your workplace, hence why it is key to address the fundamental differences between these two methods to justify their suitability for your needs.
Accuracy and Reliability
Addressing the statistical accuracy should be your first priority. The key difference between sensors and manual counts is the type of data obtained: continuous and discrete, respectively.
Manual counts are conducted by personnel, external or internal to the organization, doing timely walkthroughs around the office counting (usually helped by mobile apps or spreadsheets) the number of occupants in all the spaces agreed during the planning phase. These observers also can take additional notes about the different activities people are doing at the time of the observation, for instance: having a phone call, having a meeting and working at the desk. The walkthroughs happen every hour during the working journey (i.e. 9 am to 6 pm).
Occupancy sensors are usually installed in different spaces and work settings, such as under desks, chairs or ceiling. They record data continuously and some of their algorithms even filter false recordings such as cleaning activities done by the facilities personnel. The deployment of sensors should also be accompanied by an observational study to identify general trends in the working patterns.
In principle, conducting a round of manual counts every hour of the working journey should be reasonable to quantify issues for which probably you already know the answers, such as: identifying spaces that are never used or others that are on high demand. Nonetheless, manual counts fall short in precision and in statistical accuracy required for workplace transformation and real estate optimization.
In my experience helping users in locations such as Singapore, Hong Kong, India and collaborations with colleagues involved in projects in Europe and North America has revealed that, on average, approximately 50 percent of times where an occupant is sitting at a desk or at a meeting room will spend less than 15 minutes before leaving the seat. That figure goes up to 75 percent of times in which any occupant sitting will take no more than 45 minutes before leaving. Hence, deploying sensors has brought strong evidence to suggest that, if for manual counts done every hour, we could lose over 75 percent of all the activity that actually happened in the spaces analyzed. Moreover, conservatively assuming that each manual count takes less than 1 minute per space, we would be observing less than 1.7 percent – or losing 98.3 percent – of the total activity that actually happened during that day. So, try to guess the image of a hundred pieces jigsaw using only two of them.
In any project, there will always be time constraints, but it is very important to be aware that when it comes to data collection the longer you are able to collect it the more reliable it will be. So far, in the Corporate Real Estate (CRE) industry and Workplace Strategy field, the common practice has been to collect data during two weeks; timeframe that probably has been defined by resources’ availability, cost and timeline requirements.
From a statistical point of view, a two-week sampling period makes sense, assuming that each working week is a full cycle of the office’s activity and that the rest of the year will behave in a similar way. Sampling two weeks also allows us to determine the statistical variability and to describe how spread-out the two cycles of data are.
Nonetheless, when it comes to occupancy sensors, the sampling periods could be four weeks or more. The labor used to deploy, test and activate the sensors and the pricing schemes used by most of the suppliers justify deploying them for more than just two weeks.
Both, manual and sensor-based, methods require thorough planning before execution or deployment. The complexity lies in guaranteeing the best quality of data collected whereas causing less disruption to the business and occupants.
Manual counting requires disciplined, well-presented and reliable personnel to walk around the facilities every hour. In some cases, depending on the human resources’ availability, internal users could also collect data after receiving quick training. In this case, the major disruptions are likely to come from situations in which the observers open the door of enclosed meeting rooms or private offices to take note of the number of occupants. These situations can generate discomfort of the staff, disrupt important meetings and compromise confidential information.
On the other hand, sensors can be much less disruptive. Nonetheless, if they are deployed without any previous communications to the employees explaining the reasons and their main characteristics, they could feel they are being tracked and their privacy is being compromised, which ultimately could jeopardize the proper delivery of the project.
Even if the planning and communications are done properly, there will be a period of adaptation of the staff to the sensors. In my experience, allowing a period of adaptation of approximately one week is also important in case their natural patterns of work are affected by the impulse to remain on their work points for longer periods of time than usual.
The deployment of sensors also has to consider their collection upon the conclusion of the study, as well as their connectivity to internal or external networks. Usually, the deployment has to be completed during a weekend or after office hours. If the number of work points to be analyzed is too large, over a thousand, the complexity of the deployment will grow exponentially hence the potential need for implementing alternative strategies of sampling. On the other hand, if the sensors are located in chairs or mobile pieces of furniture, it is important to keep close track of them as they could be switched to other spaces or even other floors. This is also a reason why short-term deployments, of no more than one month, are also more effective than long-term ones.
Weighing Your Options
Currently, the cost of the sensors has decreased considerably and the algorithms have become smarter; and these trends will become stronger in the upcoming years. However, the cost is still a reason for choosing manual counts over occupancy sensors. Nonetheless, it is extremely important to be aware of the fundamental differences in the quality of data collected and how we can use it to make decisions.
If done correctly, manual counts will allow you to prove a hypothesis of limited complexity; things that you may already know, such as that corner in the office nobody uses or that booth in the other corner that is always busy. However, to conduct complex analysis, support decision making in major transformations and optimize real estate, sensors are the most reliable tool available due to their inherent characteristic of continuous data collection.
Nice article, but I believe you’ve made some incomplete assumptions:
1. There are just 2 mechanisms for measuring utilization – sensors or manual. There is a very good third option, and that is to leverage the technologies that is currently deployed in the workplace to measure use. This can be an automated connection to the physical access control system or the network to the workplace management platform like AgilQuest’s Forum.
2. The next assumption is that the period of measurement of 2 weeks or 1 month is statistically viable for all use cases. That is not the case. AgilQuest has been measuring actual use of space for 15 years, and this only works if you want to get a quick snapshot for the purposes of starting your workplace plan, seeing what’s possible for a worker to workplace ratio, etc. It’s good for that. For anything else, it is just, for all practical purposes. too short. The best mechanism for measurement is continuous, consistent and systematic over time. Measuring for 1 month in the summer, or one month in the winter, or one month at just about any other time will receive the same response, “Of course the utilization is low, it’s (fill in the blank) Summer, Winter, Spring Break, Slow Season, etc.”
Leveraging currently deployed technology via a real time integration to the management platform has the following benefits above either sensors or manual.
– The period of measurement can be continuous, consistent and systematic. This means that the client removes time of year or business cycles from the data. It also produces trends over time that you cannot receive with short term studies.
– Data becomes usable in two ways: Operational and Strategic. The operational aspect of utilization data is almost always overlooked. Everyone focuses on what happened in the past for strategic space planning, but forgets that the data has immediate operational value. For instance, a conference room is booked by a person, but they don’t come into work (don’t card in)…that room should be released for others to use. This is the main cause for the statistic that only 30 to 40% of all conference rooms that are BOOKED are ACTUALLY USED. Few things are more frustrating than walking down a long line of rooms, seeing them all booked but no one occupying them. How much money has been spent on new buildings because of this? $billions…
– The high cost of either manual or sensors (to deploy and to maintain over time) only allows them to be used for short periods of time and on small sample sizes (month or two, a floor or two). Leveraging PACs or networks allows the ENTIRE portfolio to be measured every day, now and into the future. Think of it – No company would sample their financial data from a 2 week period in just one department, then build their quarterly financials (P&L, BS, CF) for the entire company and report that to Wall Street (other than being illegal, it’s not usable) That’s what happens with small sample sizes and short duration measurements.
There is much more to discuss, but I’ll leave it at that for now. Thank you for reading my response!