Too much real estate data is not always a good thing. Data is dangerous when misunderstood.

As an Industrial Engineer, I love data and all the things I can do with a good, high-quality data set. As a Real Estate professional, I am completely used to working in a world with limited, fragmented data. These two worlds actually come together surprisingly well historically because understanding what to look for in a data set can support processes for adding in key missing bits along with better extrapolation/interpolation. However, the pandemic has caused a change [note: I am really tired of writing that phrase but unfortunately it continues to need to be reinforced] that leads to good data being used in bad ways.

One of the hardest aspects of data analysis for people who do not do much is that all data comes with unwritten context. For real estate data, that context is that the way people used offices before the pandemic is not at all reflective of how they use offices today. Similarly, I am highly skeptical that the way offices will be used two years from now will look like how they are used today.

Sure, if all you are doing is sayings we had X% of people in our offices in 2019 and we have Y% in our offices today, nothing too dangerous is happening. But there is more danger in that statement than it may seem on the surface. Let’s start with some questions about the nature of a workplace average:

• Is the average Monday through Friday? If yes, most companies have seen the most change to their occupancy on Mondays and Fridays. It is entirely likely that including both in the average is causing the before and after data comparison to look more stark than it would otherwise appear for just the middle of the week.
• Is the average across certain hours of the day? For companies that have sensors in their office, choosing the hours of the day to run an average on is a surprisingly difficult decision. In my experience, most offices experience their moment of highest occupancy right before or after lunch. However, post-pandemic the number of people in an office only working a partial day at the site can lead to the context of this average being very different now than before.
• Are you focusing on desk utilization? In 2019, the most common measure of office utilization across companies was how many desks were occupied. There were lots of ways this was done, but the “butts in seats” metric was almost universally desk focused. In 2022, people are coming to offices for a lot of different reasons, but the one reason that has decreased the most is to sit at a desk and do solo work. Collaborative activity uses fewer desks per person and can lead to a wrong view of how utilized an office is. Conference rooms are being used at higher rates than before when compared to the total number of people ocming into the office.
• Are you believing the data today is reflective of tomorrow? This is the last one and the most dangerous one. It is incredibly easy to say, “I’ve got a full year of data on my offices, it must be good and reflective of something.” I hate to be the one to break it to you, but it is not. People showing up at an office is mainly reflective of people making a decision to work from an office. If decisions to work from the office are depressed for non-office related reasons (e.g., cost of living, commute challenges, COVID fear, amenities around the office are closed, etc.) then the data is not reflective of actual future office demand.

And this was just for a simple metric calculation. Imagine how much more danger there is in a more complicated metric with less reliable data.

We all want to make data-driven decisions. Good data can help minimize the risk of our decisions being wrong or having unintended consequences. It is important to understand what “good” means though because data context is critical to a data-driven decision. Good data with a misunderstood context is equivalent to False data.