Facility managers don't need more AI. They need data they can actually use.

AI is now genuinely good at generating facility management data. Predictive maintenance signals, energy optimization numbers, space utilization patterns, that part of the promise is real. The problem is what happens after the data arrives.

What the data shows

• AI-in-FM pilots exploded from under 5% of companies to 88-92% in a few years (JLL, cross-industry pattern also seen in FM specifically)
• Many facility managers report lacking the expertise to interpret and apply AI-generated insights (IFMA "Gamechanger")
• FM data maturity is held back mainly by legacy systems, inconsistent platforms, and governance gaps, not AI capability (IFMA "Rise of the FM Analyst," 37 interviews across 6 regions)
• 60%+ of real estate companies report duplicated or dormant technology systems; 81% report at least three systems underperforming expectations (JLL)
• Appraisers cite document-gathering and data reconciliation, not judgment calls, as the most frustrating part of the job (CBRE)

More data. Not more useful.

IFMA's "Gamechanger" research on AI in facility management found something worth sitting with, because it cuts against the usual AI-adoption narrative: the technology genuinely works at the data-generation stage. Predictive maintenance signals, energy optimization recommendations, space utilization patterns, AI is good at producing all of it.

The gap is on the other side. Per IFMA's own follow-up research, many facility managers don't have the expertise, or the surrounding systems, to actually interpret and apply what the AI hands them. Getting more data was never the hard part of the job. Doing something useful with it, reliably, at scale, is.

The gap is older than the AI question

IFMA's "Rise of the FM Analyst" study, based on interviews across six global regions, traces the real constraint back further than any specific AI tool. Facility management's data maturity, as a discipline, is held back primarily by legacy building systems, inconsistent platforms across a portfolio, and low digital literacy, not by a shortage of tools or by AI capability specifically.

Even in buildings with genuinely modern systems, the same research found the basics were frequently missing: consistent data classification, systems that actually integrate with each other rather than sitting in isolation, and staff trained to work with what those systems produce. Layer AI on top of an environment like that, and the result is more signals arriving into a structure that already couldn't act on the signals it had before AI showed up.

This matters because it reframes the usual "should we adopt AI in facility management" question. The honest prerequisite question is "can our current systems and people actually use more data than we're generating today," and for a large share of the industry, per IFMA's own research, the answer is no, independent of whether AI is involved at all.

The everyday version of this problem

JLL's cross-industry technology survey adds a concrete number to the systems side of the gap: over 60% of real estate companies report duplicated functionality or dormant technology systems somewhere in their stack, and 81% report at least three existing systems that aren't delivering the results expected of them when they were purchased. That's not a hypothetical inefficiency. It's the everyday reality most facility teams are already managing around before any AI conversation even starts.

CBRE's commentary on the valuation side of the industry captures the human, unglamorous version of the same pattern. Asked what actually made the job hard, an experienced appraiser didn't point to difficult professional judgment calls, the kind of work that's genuinely hard to automate. The frustrating part was the hours spent gathering documents, chasing down inputs from different parties, and reconciling data that should have already been clean and centralized well before it reached them.

That's the consistent pattern across all of this research: the hard part of facility and property work was rarely a shortage of intelligence, human or artificial. It was the administrative grind of finding, verifying, and organizing information that already existed somewhere, in some form, before anyone, or anything, could act on it.

What actually closes the gap

IFMA's own Circular FM research, based on 36 in-depth interviews with built-environment experts, states the fix in almost understated terms: digital tools that consistently track resource use and building information aren't a nice-to-have layered on top of facility management. They're the groundwork that makes everything else, including AI, actually usable.

That has a specific, practical implication. Building information, appliance records, and warranty documentation that are organized once, kept current, and structured consistently across a portfolio solve the underlying problem before an AI ever has to compensate for it. Skip that step, and every AI tool layered on top inherits the same fragmented, dormant, duplicated foundation JLL and IFMA both describe, and produces the same disappointing results.

Frequently asked questions

Should facility teams pause AI adoption until their data is fully organized?

Not necessarily pause entirely, but sequence it correctly. Piloting AI on top of the worst-organized part of the portfolio typically produces the worst results and the most skepticism about AI generally. Starting with a well-structured subset while fixing the rest in parallel tends to build the confidence and internal case needed to expand further.

Is this mainly a large-portfolio problem, or does it affect smaller property managers too?

The proportional impact is similar regardless of size. A smaller portfolio has fewer total systems to reconcile, but the same core issue, records that aren't centralized, current, or consistently structured, produces the same daily friction, just at a smaller absolute scale.

What's the difference between "more data" and "usable data" in practice?

More data means more sensors, more reports, more numbers generated. Usable data means that data is structured, labeled consistently, connected to the right property, unit, or system, and current enough to trust without independently re-verifying it every time. IFMA's research is specifically about organizations that have plenty of the first and not enough of the second.

How do you know if your facility data is AI-ready?

A practical test: can a facility manager pull up a specific appliance's warranty status, maintenance history, and current documentation in under a minute, without calling someone else to confirm it's the right version? If not, that's the gap to close before layering AI-driven insights on top.

---

CroVIN organizes building, appliance, and warranty records once, consistently across a portfolio, so facility data is structured and current before any AI-driven insight has to compensate for a gap underneath it.