A rooftop extract fan was holding its commissioned airflow of 1.6 m³/s, with no complaints logged against it. To hit that number, it was drawing about 2.8 kW, nearly double the power the same airflow was originally designed to need.
‍

That fan wasn't on this year's capital replacement list. Would it have been, if anyone had looked? Almost certainly not. The airflow was within specification, there was no fault logged, and nothing about it looked broken. What it was quietly costing to run had little bearing on the lifecycle decision, since operational performance doesn't always feed back into the assumptions behind a capital plan. What that gap looks like, and what closed it in this case, is worth walking through.
‍

What Assumptions Actually Sit Behind a Capital Plan?

Capital plans run on assumptions: how long an asset should last, when it's likely to need replacing, roughly what that replacement will cost. Those assumptions are necessary, but useful life is a planning assumption, not a guarantee of economic life. An asset can be well within its expected lifespan and still cost meaningfully more to run than the plan accounted for, with no automatic mechanism for a fixed schedule to notice on its own. A capital plan that leans heavily on expected useful life can be slow to pick up a change like this unless current asset performance gets folded back into the decision.
‍

How Often Do Organisations Actually Check Real Asset Conditions?

According to Facility Management Journal, citing the 2026 Asset Lifecycle Report from Asset Management Software, Siemens, based on survey data from 400 qualified facilities managers across the U.S., 97 percent of organisations have a 3–5-year capital plan in place, while 32 percent have performed a Facilities Condition Assessment. An FCA, by its nature, provides an assessment at a particular point in time. It therefore complements rather than replaces continuous visibility into how an asset behaves between assessments.
‍

McKinsey's research on advanced analytics in infrastructure adds a related point, that asset owners typically already hold substantial data on the condition, maintenance, and operation of their infrastructure, and using that data well can meaningfully improve capital-planning decisions. That's a different claim than the FCA figure above; McKinsey's argument is about data being underused, not specifically about assessment cadence. But read together, they point toward the same underlying issue: the constraint usually isn't a lack of data, it's whether that data gets used in the decision at all. 
‍

Where Does That Operational Data Actually Live?

The relevant data is often already being generated, just not in one place or in a form that's easy to act on. BMS platforms typically hold continuous operating data, things like power draw and airflow over time, while maintenance records tend to hold service and work-order history rather than a live read on current condition. Having both isn't the same as turning them into a lifecycle signal, and a continuous stream sitting in a dashboard doesn't become a decision on its own. Not every organisation's data is complete or clean enough to read easily even when it is being collected.
‍

Where Does AI Actually Fit Into CapEx Decisions?

This is where AI can play a practical role in facilities management. It isn't making the capital-planning decision. It's helping generate the evidence that lets the existing decision-making process account for how an asset is actually behaving, rather than relying only on what the last assessment recorded. One application of this is autonomous maintenance: continuously interpreting operating data, identifying meaningful deviations from an asset's expected operating point, and surfacing those that warrant investigation. This doesn't require an overhaul of the entire capital-planning process; it can begin by putting the operational data an organisation already generates to work in smaller, targeted decisions.
‍

What Did This Look Like for One Rooftop Fan?

The fan mentioned earlier, from a portfolio Xempla manages, is a useful example of what this looks like in practice. Despite no complaints logged against it, the system flagged that it was running at 100 percent fan capacity to maintain its commissioned airflow of 1.6 m³/s, against an original design that expected the same airflow at roughly 80 percent capacity. That sustained maximum-capacity operation was a signal worth investigating, not proof on its own of a fault. Technicians inspected the fan on multiple occasions, and each inspection confirmed there was no mechanical issue. With a mechanical cause ruled out, the sustained inefficient operation was attributed to ageing. The fan was consistently drawing approximately 2.8 kW at 100 percent capacity, well above what its design point called for at the same airflow. Prior to replacement, the unit was estimated to be consuming approximately 32.8 kWh a day at that elevated condition, representing roughly 12,000 kWh of excess energy consumption over a year. The system surfaced the deviation early enough for technicians to investigate and establish the cause, and that evidence, together with the measurable energy penalty, supported a lifecycle replacement rather than continued monitoring. Following replacement, the new fan holds the same airflow at approximately 84 percent capacity, and the historical energy penalty was eliminated while the required airflow was maintained. 
‍

What Does the Energy Figure Actually Mean for the CapEx Decision?

That 12,000 kWh figure isn't, by itself, a replacement business case. At an actual energy tariff, that consumption could be translated into an annual operating-cost impact, but that alone still wouldn't establish payback without knowing the replacement cost and the asset's remaining useful life. What the continuous evidence changed was the quality and timing of the decision. Without it, this fan could easily have kept running at 100 percent capacity until a scheduled assessment, a failure, or some other trigger eventually brought it to attention. With it, there was measurable evidence that the fan's operating behaviour had already moved away from its expected point, which is what supported the replacement decision.
‍

What Does This Mean for a Broader Capital Programme?

One fan doesn't change a capital programme, but the same gap, an asset performing to its design output while quietly costing more than it should, can repeat unevenly across a portfolio. Individually small corrections like this one compound at scale. The point isn't to replace more assets. It's to put capital where it's actually warranted, using current asset performance alongside the lifecycle, condition, and financial factors already informing the decision.

A lifecycle plan tells you when an asset is expected to need replacement. Asset performance tells you whether waiting still makes economic sense. 

If you're curious how your own portfolio's planning assumptions compare to what your assets are actually doing, that's a conversation worth having.

‍

FAQs

Should CapEx decisions be based on an asset's age or its actual performance?

Ideally on both, but most capital plans lean heavily on age and expected useful life. Useful life is a planning assumption, not a guarantee of when an asset stops being economical to run. An asset can be within its expected lifespan yet cost meaningfully more to operate than planned. Feeding current performance into the CapEx process helps identify when that assumption may no longer make economic sense.
‍

Is a Facilities Condition Assessment enough to catch asset performance problems?

Not on its own. An FCA provides a snapshot at a point in time, so it can miss gradual changes in how an asset performs between assessments. Continuous operational data provides visibility into how an asset behaves over time, while the FCA provides periodic condition evidence. The two complement each other rather than one replacing the other.
‍

If operational data already exists, why isn't it used in capital planning?

Mostly because it often lives across different systems and isn't always in a form that's easy to act on. BMS platforms typically hold continuous operating data such as power draw and airflow, while maintenance records capture service and work-order history. Having both doesn't automatically create a lifecycle signal. The data needs to be interpreted and connected to the decisions it can inform.
‍

How can AI support CapEx decisions in facilities management?

AI doesn't make the capital-planning decision. It helps generate evidence that allows the existing decision-making process to account for how an asset is actually behaving, rather than relying only on its last assessment. In practice, AI can continuously interpret operating data, identify meaningful deviations from an asset's expected operating point, and surface those worth investigating. This gives teams more current performance evidence to consider alongside lifecycle, condition, and financial factors.