Predicting failures before they happen is no longer the difficult part. For many FM organisations, the technology already exists to identify patterns, forecast failures and generate increasingly accurate maintenance insights. Yet prediction alone doesn't create value.
The real challenge begins after the prediction has been made. Can an organisation turn that insight into timely action? As predictive maintenance matures, that question is becoming far more important than the technology itself.
Most maintenance is still reactive
Despite the industry's focus on digital transformation, maintenance in many buildings still begins when something goes wrong. An asset fails, an occupant reports an issue, or an inspection uncovers a problem that has been developing unnoticed for weeks.
The reasons are usually the same: fragmented data, disconnected systems, incomplete maintenance logs and limited budgets continue to prevent organisations from taking a more proactive approach.
Research consistently shows that this remains the operational reality across much of the industry, even among organisations that consider themselves digitally mature.
That's why the journey towards predictive maintenance is about far more than implementing new technology. It starts with building reliable data foundations that organisations can trust and act upon.
Understanding the maintenance maturity journey
Predictive maintenance didn't emerge overnight. It is the latest stage in the evolution of maintenance, with each step building on the last. The progression itself is straightforward. Where many organisations go wrong is assuming they've reached the final stage simply because they've installed sensors and dashboards.
Reactive maintenance is run-to-failure. An asset stops working, someone responds, and normal operation is restored. More organisations spend time in this stage than they would probably care to admit.
Preventive maintenance introduces a schedule. Calendar-driven. Interval-based. It reduces unexpected failures but creates a different challenge: components are often serviced or replaced simply because the schedule says so, not because they actually need attention.
Condition-based maintenance changes the trigger. Instead of relying on the calendar, maintenance is based on the actual condition of the asset. Measurements and monitoring determine when intervention is required. This is where many organisations believe they've reached predictive maintenance after installing sensors and dashboards. In reality, they often haven't.
Predictive maintenance is fundamentally different. Rather than identifying problems as they happen, it forecasts when an asset is likely to fail by analysing changes in its condition over time. A sensor telling you that a motor is overheating today is condition monitoring. A system predicting that the motor is likely to fail within the next three weeks based on performance trends is predictive maintenance. The distinction may seem subtle, but confusing the two can lead organisations to overestimate the maturity of their maintenance capabilities.
Technology isn't the hard part
It's easy to think that the journey to predictive maintenance is primarily about choosing the right technology. In practice, the challenge is usually much broader.
Predictive models depend on reliable data. Yet many organisations are working with incomplete maintenance histories, disconnected building systems, inconsistent asset information and valuable operational knowledge that never makes it beyond the technician who discovered it.
Research consistently shows that the biggest obstacles are not the predictive algorithms themselves, but the complexity of bringing together data from different sources and translating research models into real operational environments. Technology continues to advance rapidly. The challenge is ensuring organisations and their data are ready to take advantage of it. Preparing the organisation and its data to make effective use of that technology is often where the real work begins.
Choosing the right technology is an essential part of the journey. But technology alone isn't enough. Predictive maintenance depends just as much on reliable data, connected systems and the organisational foundations needed to turn insights into action.
Start where the impact is greatest
For most organisations, HVAC is the logical place to begin. It is one of the most energy-intensive systems in a building, operates almost continuously and has a direct impact on occupant comfort, indoor air quality and business continuity. When it underperforms, the consequences are felt quickly, both operationally and financially.
Research suggests that inefficient HVAC systems can increase energy consumption by 20–30%, making even modest improvements capable of delivering measurable value.
Rather than attempting to predict failures across an entire portfolio from day one, a more practical approach is to focus on a small number of critical assets. This allows organisations to establish reliable data, validate predictive models and refine operational processes before scaling more broadly.
From prediction to action
Predictive maintenance answers an important question: What is likely to happen? Prescriptive maintenance takes the next step by asking: What should we do about it? It considers factors such as technician availability, maintenance schedules, operational priorities, contractual obligations and cost to recommend the most appropriate course of action.
That's an important step forward. But it still isn't the whole story.
Imagine a system predicts that a critical HVAC asset is likely to fail within the next three weeks. The prediction has been made. Prescriptive maintenance takes this prediction to the next level. It turns the prediction into a window of opportunity and prescribes appropriate action within a timeframe when the right technician and required material is available and the occupancy in the specific room is low. In both cases the technology has done its job. What happens next depends on the organisation.
Who owns the decision? Who approves the work? Which team decides whether the maintenance should take priority? What happens if operations, finance and facilities all have different priorities? None of those questions are answered by the technology itself.
The closer organisations move towards predictive and prescriptive maintenance, the challenge changes. It is no longer about whether technology can predict a failure. It is about whether the organisation is prepared to respond.
Prediction alone doesn't create value. Action does.
The next challenge is organisational
Buildings are generating more operational data than ever before. Our ability to collect and analyse that data continues to improve. What isn't keeping pace is the organisational capacity to act on what the data reveals.
The reason is simple. Technology evolves quickly. Organisations don't. Many FM organisations were designed around scheduled maintenance and reactive repairs. Their approval processes and decision-making structures were built for information that arrived periodically and could wait while people deliberated. A prediction with a three-week window can't wait. When acting on it requires a monthly planning cycle or a quarterly budget review, the organisation isn't slow. It simply wasn't designed to move at the speed of the insight.
That isn't a criticism of the people running these organisations. It's a reflection of systems that were designed for a different era. The technology has evolved. The data has evolved. Many organisations are still working within decision-making structures that haven't.
Three questions worth asking:
- For your most critical assets, do you know not only their current condition, but also when they are likely to require intervention?
- Is your maintenance history structured in a way that systems can actually learn from, or does critical knowledge still reside primarily with experienced technicians?
- Most importantly: who owns the decisions that follow the insights your buildings generate?
This is one of the challenges I explore in my upcoming book, The Acceleration Gap, where I examine why organisations are becoming increasingly capable of generating insights, yet often struggle to act on them. Because in the end, the question isn't whether your organisation has the data. It's whether it's built to act on it.