Asset performance management: what APM covers, the data it runs on, how it differs from a CMMS, and how to tell whether it is working
What asset performance management covers, the loop from collecting an asset's data to measuring the result of the work it raised, the data underneath it and the usual gaps, what APM software does and what each capability needs, how it sits beside the CMMS, historian and condition monitoring, bought or built, the checks before starting, and how to tell it is working.
Key takeaways
- Asset performance management, or APM, is the work of judging which assets need attention and why, from their condition, process data and failure history, and turning that judgement into maintenance action.
- APM software sits beside the CMMS or EAM rather than replacing it. The CMMS holds the work. APM decides which work matters, and raises it into the CMMS.
- It runs on data most sites already have but have never joined: historian tags, condition results, work order history and the asset register. Joining them to one asset record is most of the effort.
- Prediction is only as good as the failure history behind it. A site whose work orders carry no failure codes has little to learn from, whatever the software claims.
- APM works when every finding has an owner and a path into the CMMS, and when the closed work order comes back to tell the analysis whether it was right.
A site installs online vibration monitoring on its critical conveyors and connects it to a dashboard. Within a month the dashboard shows forty amber assets. Nobody knows which of them matter, the alerts do not reach the work management system, and the reliability engineer is the only person who opens the screen. A year later a gearbox fails that had been amber for six weeks. The monitoring worked. What was missing was the rest of asset performance management: the judgement of which findings matter, a person who owns them, and a path from the finding to the work.
This guide sets out what APM covers, the loop it runs, the data underneath it, what the software does, how it differs from the CMMS or EAM, the historian and condition monitoring, what to have in place before buying or building it, and how to tell whether it is working. It takes ISO 55000 as the frame, ISO 14224 for failure data and ISO 13374 for the processing of condition data.
The loop APM runs
APM is a loop with four steps. Collect brings an asset's data together against one record. Judge turns that data into a view of the asset's health and the risk of it failing. Decide chooses the response: a work request, a change to the maintenance strategy, a spare held, or a decision to defer, repair or replace. Act and measure is the work done through the CMMS, with the result recorded so that the next round of judgement knows whether the last one was right.
The software mostly adds the second step. The other three depend on the site's data, people and work management, which is why two sites with the same product can get very different results from it.
1Collect
Bring the asset's data together against one asset record
- Historian tags
- Condition monitoring
- Work order history
- Inspections and rounds
2Judge
Turn the data into a view of health and risk
- Asset health
- Failure risk
- Criticality
- Remaining life
3Decide
Choose the response, and who it goes to
- Raise a work request
- Change the strategy
- Hold a spare
- Defer, repair or replace
4Act and measure
The work is done in the CMMS and its result recorded
- Work order closed
- Failure coded
- Condition after repair
- Performance trended
↺ The result of each job is the next round's data. Without the return to step 1, APM is a dashboard
The data underneath it
APM runs on data a site usually has already, in five or six places, each keyed differently. The historian knows the tag, the condition analyst knows the measurement point, the CMMS knows the functional location, and the asset register knows the asset number. Mapping those to each other, once and in a way that is maintained, is the largest part of most APM projects. It is the master data problem again, applied to analysis rather than scheduling.
| Source | What it gives APM | The usual gap |
|---|---|---|
| Asset register and hierarchy | The asset each reading and work order belongs to, its criticality and class | Tags in the historian that do not map to assets in the CMMS |
| Process data | Run hours, loads, temperatures, pressures, starts and trips from the historian | Collected for operations, at rates and with names that do not suit asset analysis |
| Condition monitoring | Vibration, oil, thermography and ultrasound results, online or by route | Held in the analyst's own software and never joined to the asset record |
| Work order history | What failed, when, how it was found and what was done | Free text and missing failure codes, so failures cannot be counted |
| Inspections and rounds | Readings and observations from operators and inspectors | On paper, or in a form whose results go nowhere |
| Design and maker's data | Ratings, limits, recommended intervals and expected life | In documents rather than fields the software can use |
An APM project is mostly a data project. The analytics are the easy part once every reading knows which asset it belongs to.
What APM software does
Products describe their capabilities differently, but the same six appear in most of them. Each depends on something the site has to supply. Failure prediction is the most marketed and the most dependent: a model learns from failures, and a site with few recorded failures, or failures recorded in free text, gives it little to learn from. Rules written by engineers, such as a vibration level above the alarm limit for a set time on a critical asset, often deliver more in the first year.
| Capability | What it does | What it needs underneath |
|---|---|---|
| Asset health | Combines condition, process and history into a health score or state for each asset | Agreed rules for what each reading means, by asset class |
| Failure prediction | Detects patterns that precede failure, by rules, statistics or machine learning models | Enough failures, recorded well, to learn from or to test against |
| Risk and criticality | Ranks assets by the likelihood and consequence of failure | A criticality assessment the site agrees with |
| Reliability analysis | Bad actors, failure rates, MTBF and Weibull by asset and failure mode | Failure coding to a standard such as ISO 14224 |
| Strategy management | Holds the maintenance strategy for each asset and tests it against what is failing | An FMECA or RCM basis for the strategy |
| Case management | Tracks each finding from detection to the work order and the outcome | The link into the CMMS, and someone who owns each case |
Beside the CMMS, the historian and condition monitoring
APM overlaps with three systems a site will already have, and the overlap is where the confusion comes from. The CMMS or EAM manages work. The historian stores what the plant did. Condition monitoring measures what state a machine is in. APM sits across them and asks one question the others do not: which assets need attention, and why. Its output is a decision, and the decision has to land in the CMMS, or it does not happen.
| APM | CMMS or EAM | Historian | Condition monitoring | |
|---|---|---|---|---|
| Main question | Which assets need attention, and why | What work is done, by whom, with which parts | What did the plant do, second by second | What condition is this machine in |
| Holds | Health, risk, findings and strategy | Work orders, history, plans, parts and costs | Time-series process data | Measurements and the analyst's diagnosis |
| Acts by | Raising work in the CMMS or changing a strategy | Planning and scheduling the work | Feeding other systems and people | Reports and alerts to the reliability team |
| Without the others | A dashboard of scores nobody acts on | Records work without knowing which work matters | Data without an asset or a decision | Findings that never become work orders |
Bought or built
APM can be a product or an arrangement of parts. A bought suite brings a tested set of analytics, a user interface and a data model, and suits a site that can adopt the product's way of working. Built on the site's own historian, CMMS and an analytics layer, it can follow the site's existing processes and asset structure, at the cost of the site owning the software. Many sites end up between the two, with a product for the analytics and integration work to connect it.
Either way the work that decides success is the same: the asset mapping, the failure coding, the rules for what each reading means, and the path from a finding to a work order. The system integration guide covers the interfaces involved. Doing that work first is also the cheapest way to find out what the site actually needs from the software.
Before buying or building
Six checks show whether a site is ready for APM, and each gap is cheaper to close before the software arrives than after. None of them is about the software.
| Check | Ready when |
|---|---|
| Asset register | Every critical asset exists once, with a criticality, and its historian tags and condition points are mapped to it |
| Failure history | Work orders carry a failure code and a cause, so failures can be counted by mode |
| Condition data | Condition results are in a system, against the asset, not in reports |
| A decision owner | A named person reviews findings and decides on each one within an agreed time |
| The return path | A finding raises a work request in the CMMS, and the closed work order comes back to the case |
| A measure | The site has chosen what APM should improve, such as unplanned downtime on critical assets, and has a baseline |
How to tell it is working
APM is working when findings turn into work and the work changes the plant's results. Useful measures follow that chain: the share of findings reviewed within the agreed time, the share that become work orders, the share of failures on monitored assets that were detected before they happened, and the trend in unplanned downtime and maintenance cost on critical assets. The maintenance KPIs guide covers the downstream measures.
Measures of activity, such as the number of assets connected, models deployed or alerts raised, say only that the system is busy. A rising alert count with a flat failure rate usually means findings are being generated faster than anyone is acting on them.
Where APM goes wrong
- Software bought before the asset register, failure coding and condition data are in order.
- Health scores shown on a dashboard with no owner and no path into the CMMS.
- Predictive models promised on a failure history too thin or too poorly coded to learn from.
- Every asset connected at once, instead of the critical ones first.
- Alert thresholds left at defaults, so the alert count grows until people stop reading them.
- No return path, so the analysis never learns whether its findings were right.
- Success measured by how much is connected rather than by what failed less.
Common questions
What is asset performance management?
Asset performance management is the practice, and the software that supports it, of monitoring the health and risk of physical assets and deciding what to do about them. It brings together condition monitoring results, process data from the historian, work order history and the asset register, judges which assets are degrading or at risk, and raises maintenance action. Its aim is to improve availability and reliability at the least cost and risk, in support of an asset management system such as ISO 55001.
What is the difference between APM and a CMMS?
A CMMS or EAM manages the work: work orders, plans, schedules, parts, labour and history. APM manages the judgement about which assets need work and why: health, risk, predicted failures and the maintenance strategy. A CMMS records that a pump bearing was replaced. APM noticed the bearing degrading and raised the request. Most sites need the CMMS first, because APM depends on its history and acts through it.
What does APM software do?
Typical capabilities are asset health scoring, failure prediction by rules, statistics or machine learning, risk and criticality ranking, reliability analysis such as bad actors and failure rates, management of the maintenance strategy for each asset, and case management that tracks each finding from detection to the work order and its outcome. Products differ in emphasis, and the same capabilities can be assembled from a historian, a CMMS and analytics tools.
Is APM the same as predictive maintenance?
No. Predictive maintenance is one part of APM: using condition data to forecast a failure and act before it happens. APM is wider. It also covers assets with no condition monitoring, the risk ranking that decides where monitoring is worth it, reliability analysis of the history, and the review of maintenance strategies against what is actually failing.
What data does APM need?
An asset register with criticality, process data from the historian mapped to those assets, condition monitoring results against the same assets, and a work order history with failure codes and causes. Design and maker's data, such as ratings and limits, help set the rules. The hardest part is usually not collecting the data but joining it, because each source names the asset differently.
Should APM be bought or built?
Either can be right. A bought suite brings tested analytics and a ready user interface, and suits sites that can adopt its data model and workflow. Built from parts, on the site's historian, CMMS and an analytics layer, it can fit existing systems and processes more closely and avoid another platform to license and support. The deciding factor is usually the data and the decision process, which have to be in place either way, rather than the software.
Key terms
Plain-language definitions from our glossary for the concepts this article leans on.
Standards and further reading
- ISO 55000:2024 Asset management, vocabulary, overview and principles (ISO)
- ISO 55001:2024 Asset management system requirements (ISO)
- ISO 14224:2016 Collection and exchange of reliability and maintenance data for equipment (ISO)
- ISO 13374-1 Condition monitoring data processing, communication and presentation (ISO)
- ISO 17359:2018 Condition monitoring and diagnostics of machines, general guidelines (ISO)
Related case studies and tools
- Predictive maintenance and condition monitoring (case study)
Related reading
Working through something like this?
See how we approach these initiatives, or tell us what you are dealing with.