A practical framework for facility data readiness

ABM's What's Possible expert panel series recently hosted a conversation about AI and the future of facility services with Kayla Oliver, Head of Products, Partnerships, and Innovation, and Bob Clarke, who leads Client Experience and Operations Support. A key question Oliver and Clarke kept returning to: what does a facility need to have in place before AI is useful?

AI hinges on data. Yet, for most facility managers, data quality and integration are the biggest barriers to scaling AI, outranking budget constraints, cybersecurity concerns, and lack of expertise.

Maintenance logs live in one system, occupancy counts live in another, and neither one talks to the building management system. Some of it is incomplete. Some of it was never digitized in the first place.

Every conversation about AI seems to assume a level of data maturity most facilities have not reached yet. That gap stops a lot of people before they start. But, as our experts discuss, the place to begin is not a platform or a data strategy. It is something closer to a habit.

In part 3 of this recap series, we look at how facility leaders can improve data readiness, on a small scale, and with limited resources.

Start with what you know

Bob Clarke uses an example from his first job collecting data, long before facilities had anything to do with it. "I used to be an usher in my church," he said. "When I was a kid, part of my responsibility was carrying a clicker. Every Sunday we would go around and click how many people were in the church, because we wanted to see how many people were coming in."

His point is that data collection does need to be sophisticated to be useful. "The data collection process is as old as time," Clarke said. "We overcomplicate things."

Facility leaders who feel behind can start small. Count people. Count assets.

"How many people do you have in the building who are employees? Depending on your facility type, how many people do you have coming into the building?" he said. "That is the basic principle."

The same logic applies to equipment. "Different organizations define assets differently, but think about the assets that really operate the facility, whether that is HVAC units, lighting, fire extinguishers, rooftop units, or exit signs," said Clarke. "You can boil the ocean in terms of assets. Pick the core assets that are critical to the operation of the facility. That is the baseline for a good AI implementation. It is not that complex, but you have to be curious about what you have."

Follow a three-part framework

Once a facility has that baseline, the work shifts from collecting to connecting. Kayla Oliver breaks that next stage into three moves.

First, give every system a shared reference point, typically a facility or asset number, so a reading in the building management system and a work order in the CMMS can be tied to the same thing.

Second, connect the underlying systems to each other. A shared ID means nothing if the platforms still cannot talk to one another.

Third, once the basics are unified, expand what gets measured. Add occupancy data or other IoT inputs where they are missing. "IoT is getting incredibly cheap while AI is getting increasingly powerful, and those two things converge to be even more powerful," Oliver said, "because IoT gives you more data, and AI makes that data more powerful."

The order in which facility managers apply this framework matters. Facilities that reach for sensors and platforms before they have a basic count in place tend to end up with more data and no better decisions, which is the same trap as jumping straight to AI without doing the groundwork underneath it.

Imperfect data is not a reason to wait

When it comes to AI readiness, perfection is the enemy of progress. "AI is getting better and better. Even if you do not have all the data, that does not mean you should not start," said Oliver. "People wanting AI for the sake of AI is not the right place to start. The right place to start is data and outcomes."

Be thoughtful, follow good practices, get the data in order, but do not wait for perfect. Waiting is how facilities never start.

In the final part of this series: counting people and assets is only useful if someone knows what to do with the numbers. Part four looks at the workforce side of AI readiness. We’ll examine what facility teams need to learn, and how their roles change, once AI is actually part of the job.