Why more data does not guarantee greater reliability

Every modern manufacturing plant generates more asset data than ever before. Sensors continuously measure vibration, temperature, pressure, flow and lubrication, while connected equipment streams operational information from across the factory floor. Yet despite unprecedented visibility into machine performance, many manufacturers continue to suffer avoidable failures, unplanned downtime and costly maintenance interventions.

The challenge is no longer collecting information. It is understanding which information matters, who needs to act on it and how quickly decisions should be made. According to Drew Mackley, Director of Sales Enablement for Reliability Solutions at Emerson, that shift is redefining reliability from a maintenance activity into an operational capability that influences productivity across the entire enterprise.

For many years, reliability initiatives focused primarily on detecting equipment faults before they developed into failures. Today, manufacturers have access to far more sophisticated monitoring technologies, AI-powered analytics and enterprise-wide visibility than ever before. However, Mackley argues that more information does not automatically lead to better operational decisions. In fact, providing the wrong information to the wrong people can be just as unhelpful as having no information at all.

“Operations only need information they can actually act on,” he explains. “You don’t want to put an alert on somebody’s screen simply for the sake of putting an alert there. If a bearing is beginning to fail, maintenance is usually the team that needs to respond. But when you’re dealing with critical assets where operators can reduce load, adjust operating conditions or take other corrective actions, then sharing that information with operations becomes valuable.”

That distinction reflects a broader change taking place across manufacturing. Reliability is no longer measured simply by how effectively maintenance teams identify equipment problems. Increasingly, it depends on how well manufacturers translate continuous streams of asset intelligence into decisions that improve availability, protect production and use increasingly scarce engineering expertise more effectively. The competitive advantage comes not from generating more maintenance data, but from ensuring the right people receive the right insight at precisely the moment they can still influence the outcome.

Better decisions begin with better data

AI is becoming an increasingly important tool in asset reliability, but Mackley believes manufacturers need to be realistic about what it can and cannot do. While many technology providers present AI as a fully autonomous solution capable of monitoring equipment and diagnosing faults without human intervention, he argues that the quality of any recommendation will always depend on the quality of the underlying data.

“AI is only as good as what it’s been taught,” he says. “It doesn’t create or reason in the way people sometimes imagine. It recognises patterns based on what it has already seen. If your sensing technology isn’t collecting the right information, no amount of AI is going to identify a fault that simply isn’t being measured.”

The example Mackley gives is deliberately straightforward. If a vibration sensor cannot measure the frequency at which a gearbox fault occurs, AI cannot compensate for the missing information. The limitation is not the analytics but the quality of the data entering the model. As he puts it, the familiar principle of ‘garbage in, garbage out’ still applies, regardless of how sophisticated AI becomes.

That does not diminish AI’s value. Instead, Mackley sees it as an efficiency tool that helps maintenance teams focus their attention more effectively. Pattern recognition can rapidly identify equipment behaving differently from normal operating conditions, highlight developing faults and prioritise assets requiring further investigation. However, understanding the severity of a problem, estimating remaining useful life or deciding whether an asset can continue operating safely still relies heavily on engineering judgement.

Rather than replacing experienced reliability specialists, AI is beginning to augment them. By automating routine analysis and highlighting emerging issues, it enables smaller maintenance teams to spend less time searching through data and more time diagnosing root causes, planning corrective action and preventing failures before they affect production. In Mackley’s view, that balance between automation and human expertise will define the next generation of asset reliability far more than the pursuit of fully autonomous maintenance.

For many manufacturers, the greatest obstacle to improving reliability is not a lack of data but a lack of continuity. Traditional route-based condition monitoring, where technicians collect vibration readings once a month or once a quarter, provides only isolated snapshots of machine health. Between those inspections, critical assets may deteriorate unnoticed, leaving maintenance teams to react to failures rather than manage them proactively.

Mackley believes continuous condition monitoring has changed that equation, transforming reliability from a periodic maintenance exercise into an ongoing operational capability.

“If you’re collecting data once a month, you only know what the health of that asset looked like the last time you visited it,” he explains. “A lot can happen in 30 days, particularly on critical, high-speed equipment. If the machine isn’t running when you arrive, you skip it, and suddenly you’ve got a 60-day blind spot. Continuous monitoring removes those blind spots and gives you much more time to react as conditions change.”

The shift towards continuous monitoring accelerated during the COVID-19 pandemic, when many maintenance teams were unable to access production facilities as frequently as before. Manufacturers that had relied on manual data collection suddenly found themselves making decisions with limited visibility into asset condition. In response, many accelerated investments in permanently connected sensors, wireless monitoring and cloud-based platforms that enabled engineers to assess equipment health remotely.

That evolution has also changed how reliability information is used across the enterprise. Rather than existing as isolated maintenance records within individual plants, asset health data can now be viewed across multiple sites through centralized asset performance management platforms. Engineers responsible for several facilities can compare equipment health, identify recurring issues and prioritize resources where they will have the greatest operational impact, while managers gain a clearer understanding of reliability across the entire organization.

For Mackley, this broader visibility represents the real value of digital reliability. Manufacturers no longer need to wait for the next inspection route to understand the condition of critical assets. Instead, they can begin each day with an up-to-date picture of equipment health, allowing maintenance decisions to become increasingly proactive rather than reactive.

Reliability becomes an operational capability

As monitoring technologies mature, manufacturers are also changing the way they think about reliability itself. For years, the conversation centred on predictive maintenance, identifying equipment faults before failure occurred. Mackley believes that description is now too narrow. Predicting failures remains important, but the ultimate objective is not better maintenance. It is healthier assets, lower operational risk and more consistent production.

“I think we’re becoming much more focused on the outcome,” he explains. “Predictive maintenance describes part of the process, but the real objective is understanding the health of our assets and then taking the right corrective action. There’s a workflow around that, from identifying a problem to planning maintenance and managing the risk to production.”

That broader perspective also changes how manufacturers approach different classes of equipment. Not every asset requires the same level of monitoring, nor should every machine be managed in the same way. Highly critical equipment may justify continuous, high-speed monitoring with immediate shutdown protection, while lower-risk assets can often be monitored less frequently or, in some cases, allowed to run to failure because replacement is inexpensive and operational consequences are minimal.

“The idea that every asset should be monitored in exactly the same way simply doesn’t reflect how manufacturing works,” Mackley says. “You need different strategies depending on the criticality of the equipment, how quickly conditions can change and what impact failure would have on production.”

Perhaps the most significant development is that reliability data is no longer viewed in isolation. Information from vibration sensors increasingly sits alongside process variables such as pressure, temperature, flow and valve position, allowing engineers to understand not only that a machine is failing, but why. A pump experiencing cavitation, for example, may initially appear to have a vibration problem, when the underlying cause is an upstream process issue affecting flow. Bringing these information streams together allows manufacturers to move beyond treating symptoms and instead address the root causes that reduce reliability.

For Mackley, that integration represents the future of asset health. Reliability is becoming less about monitoring individual machines and more about understanding how the entire production process behaves as a connected system, enabling maintenance, operations and engineering teams to make better decisions from a shared view of plant performance.

From information to action

Reliability is no longer defined by how much maintenance data manufacturers collect, but by how effectively they use it. AI, continuous monitoring and connected asset intelligence are making it easier to identify developing problems, but their real value lies in helping maintenance and operations teams make better decisions before failures affect production. The manufacturers gaining the greatest benefit will be those that integrate technology with engineering expertise, applying the right monitoring strategy to the right assets and ensuring that insights lead to action.

“If you’re not monitoring asset health today, the first question should be, ‘Why not?'” Mackley concludes. “The technology is now so attainable. Start with your biggest pain points, understand what information you need and choose the approach that fits your assets, your people and your operation.”

Ultimately, reliability is becoming less about predicting failure and more about creating the operational insight needed to keep production performing at its best.