By Chelsey Moir, Marketing Manager
Thirty alerts. One maintenance decision.
One exhaust fan. Around 30 alerts a week.
That was the experience of one MOVUS customer using a predictive monitoring approach. The system was doing what it was designed to do: detecting changes in machine behaviour and alerting the maintenance team when something moved outside expected parameters.
The problem was what happened next… Each alert still needed to be interpreted. Someone had to determine whether it represented a genuine problem, how serious that problem was, and what action, if any, the maintenance team should take.
After the site transitioned to PlantOS™, those alerts were consolidated into one diagnostic report that the team could act on. According to the site’s engineer, the team acted on the recommendation, helping prevent the issue from developing into a catastrophic equipment failure.
That example illustrates the difference between predictive and prescriptive maintenance better than almost any definition can.
Predictive maintenance helps teams understand what is likely to happen.
Prescriptive maintenance takes the next step by helping them decide what to do about it.
What is predictive maintenance?
Predictive maintenance uses equipment condition and operational data to identify deterioration and anticipate when maintenance may be required or when there is a risk of failure.
Depending on the application, this data may include vibration, temperature, ultrasound, oil analysis, electrical signals, process parameters, and historical machine behaviour.
Rather than maintaining equipment solely according to a fixed schedule, teams can use the asset’s actual condition to understand when intervention may be required. This has been an important development for industrial maintenance. Detecting deterioration before an asset fails gives teams something reactive maintenance never could: time.
However, a prediction still needs to be interpreted. If a system tells a reliability engineer or maintenance team that a bearing is deteriorating, they still need to decide how serious it is, how quickly it is progressing, and whether the equipment can continue operating safely.
That is where the distinction between prediction and prescription becomes important.
What is prescriptive maintenance?
Prescriptive maintenance builds on condition monitoring and predictive maintenance by helping teams determine the appropriate response to a developing equipment issue.
It can bring together information such as the probable failure mode, severity, asset criticality, and operating context to provide clearer guidance about what action should be taken and when.
Put simply, predictive maintenance tells you there is a decision to make. Prescriptive maintenance helps you make it.
Working closely with industrial maintenance teams showed MOVUS that detection was only part of the challenge. An alert may tell someone where to look, but it doesn’t necessarily explain what they are likely to find, how urgent the issue is, or what they should do next.
PlantOS was developed to help close that gap by moving the output from individual alerts towards diagnosis, prioritisation and recommended action.
Predictive vs prescriptive maintenance at a glance
| Predictive maintenance | Prescriptive maintenance | |
| Primary question | What is likely to happen? | What should we do about it? |
| Identifies deterioration | Yes | Yes |
| Anticipates developing issues | Yes | Yes |
| Diagnoses the likely cause | Sometimes (depending on the system) | Yes |
| Considers severity | Limited / varies | Yes |
| Considers asset criticality | Limited / varies | Yes |
| Incorporates operational context | Limited / varies | Yes |
| Recommends action | Generally no | Yes |
| Primary Goal | Earlier warning | Better-informed decisions and action |
The difference is not simply more sophisticated analytics. It is what happens after an issue has been identified.
What actually changes when you move to prescriptive maintenance?
Consider a drive-end bearing on a critical motor.
Basic condition monitoring might identify that vibration has increased. Predictive maintenance can take that information further and identify that bearing condition is deteriorating, and the risk of failure is increasing.
Both are valuable, but the maintenance team still has questions:
- What is causing deterioration?
- How severe is it?
- How quickly is it progressing?
- Can the motor continue operating?
- When should it be inspected, and what should the technician look for?
A prescriptive approach helps close this gap by moving from the developing signal to the likely fault, then providing a recommended next action for the responsible team to assess. This is why generating more alerts is not necessarily the same as improving maintenance.
In the exhaust fan example, PlantOS did not stop monitoring the machine. It continued analysing its condition, but instead of asking the customer to interpret around 30 separate alerts, it presented the developing issue through one diagnostic report.
The maintenance team had fewer messages to interpret and clearer information on which to base its decision.
Why does operational context matter?
Machine condition is only part of a maintenance decision.
Suppose two identical motors show the same signs of bearing deterioration. The first has a standby available, operates intermittently, and has a replacement bearing sitting in the storeroom. The second drives a production bottleneck, runs continuously and has no redundancy.
The condition signal may be similar, but the appropriate maintenance response should not necessarily be the same. This is where context becomes important.
A useful prescription needs to consider not only what is happening inside the machine, but what that issue means to the broader operation. Relevant context may include:
- The severity and rate of deterioration
- The criticality of the asset
- Current operating conditions
- Maintenance history
- Spare-part availability
- Redundancy
- The consequences of the asset becoming unavailable
Bringing this information together helps a team move from “we have a problem” towards “this is the appropriate response to this problem”.
Does prescriptive maintenance replace predictive maintenance?
No. Prescriptive maintenance depends on the ability to detect and understand developing equipment issues. Condition monitoring and predictive analytics remain important parts of that process. The difference is that the output does not stop at detection or prediction.
During our Austmine webinar, MOVUS CEO Sanjeev Kumar described this evolution as moving from alerts to recommendations, detection to diagnosis, and dashboards to decisions.
A dashboard can provide an experienced engineer with valuable information, but if that person still needs to interpret multiple trends, consult several systems and determine the appropriate response, there is still work to be done between insight and action.
Prescriptive maintenance aims to shorten that distance.
Do you need to replace your existing maintenance systems?
Moving to a prescriptive approach also does not mean replacing the systems already operating across a plant.
Most industrial sites already have valuable infrastructure in place. There may be sensors and condition monitoring systems at the machine level, PLC and SCADA data across the process, historians containing years of operating information and systems such as SAP or IBM Maximo managing maintenance activity.
PlantOS is designed to work within and complement that environment rather than requiring it to be replaced, helping organisations extract more value from the asset and operational data they already collect.
Depending on the site’s requiements and configuration, PlantOS can operate as an independent platform or exchange data with other systems through APIs. In the Austmine webinar, Sanjeev described PlantOS as sitting between the OT and IT layers, turning information from the plant floor into asset intelligence that can support the systems and teams above it.
For plants that have invested heavily in monitoring or maintenance technology, the shift towards prescription can be about extracting more value from existing infrastructure, not starting again.
What happens when you can’t act immediately?
One of the best ways to understand the value of prescription is to consider what happens when a maintenance team knows there is a problem but cannot immediately fix it.
During the Austmine webinar, Sanjeev shared an example involving a ball mill at an aluminium smelting operation. Before deployment, the operation had experienced 30 different types of failures and approximately 176 hours of unplanned downtime over a two-year period. Over the following year and a half, Sanjeev reported five different types of failures.
One of those involved in a gearbox problem identified approximately a month in advance. There was a complication… the required spare was not available on site.
The early warning did not make the supply problem disappear, nor could the team simply stop production for a month. What it did provide was visibility. The operation understood what was developing and could adjust its production plans while the situation was managed.
This is why the value of predictive and prescriptive maintenance should not be reduced to whether a machine is stopped before it fails.
Sometimes the appropriate action is an urgent inspection or repair. In other situations, it may be ordering a spare, changing operating conditions, planning work around an upcoming shutdown or continuing to operate while closely monitoring the issue.
Earlier warning creates time. Better context helps teams decide how to use that time.
How does PlantOS move from prediction to prescription?
PlantOS is MOVUS’ prescriptive AI platform for industrial asset performance. It uses a closed-loop approach that moves from detection through to diagnosis, recommendation, action and validation.
PlantOS analyses machine behaviour to identify developing anomalies. When the available evidence indicates an issue, the platform identifies the likely fault and provides a recommended action for the responsible team to assess. The maintenance team can then plan and complete the work and validate what was actually found.
That last step is important because maintenance teams bring operational knowledge and experience that does not exist in sensor data alone. Capturing what was found helps close the loop between AI analysis and what is actually happening on the plant floor.
The goal is not to remove the expert from the decision. It is to give that expert clearer information on which to make it.
The difference is what happens next
Predictive and prescriptive maintenance are not competing approaches. Prescription builds on the visibility that condition monitoring and predictive maintenance have already created.
Predictive maintenance gives teams more time by helping them understand what is likely to happen.
Prescriptive maintenance helps them make better use of that time by explaining what the issue means and recommending what they should consider doing next.
That is a relatively simple distinction, but on the plant floor it can be the difference between another alert to investigate and a maintenance decision a team can actually act on.
Already using predictive maintenance? See what comes next.
PlantOS helps industrial teams move from detecting and predicting equipment issues to diagnosing what is happening and determining the appropriate next action.