Contact

Blog

Failure codes, criticality, AI: getting more out of your CMMS data with IBM Maximo 

A CMMS, sensors, years of maintenance history: most maintenance teams now have plenty of data. What makes the difference is how that data is structured and put to work. Here are four practical levers for turning it into decisions, and how IBM Maximo Application Suite helps you put them into practice. 

Build knowledge around the asset 

When an asset’s history is scattered across the CMMS, spreadsheets and technicians’ memories, every diagnosis starts from scratch. The first rule is to make the asset record the single point of entry. Work orders, readings, manufacturer documents and spare parts should be linked to it, not just stored somewhere. 

The second rule concerns how work orders are closed. A free-text entry such as “pump repair” is useless six months later. A few structured fields at close-out are enough: symptom observed, cause identified, action taken, parts used. With those four pieces of information, you can find similar jobs and calculate KPIs you can rely on. Finally, asking experienced technicians to record their less obvious diagnoses in the work order itself is still the simplest way to hold on to their know-how when they leave. 

Failure codes technicians actually use 

An MTBF figure or an FMEA is only as good as the failure codes behind it. When “Other” dominates the statistics, the problem almost always lies in how the code lists are built, not with the technicians. 

Three principles make a real difference to data quality: 

One point is often overlooked: distinguishing the functional location (for example, “line 2 feed pump”) from the physical asset installed there. It is the only way to follow the history of a motor moved from one line to another, and to tell whether a recurring failure comes from the equipment itself or from its operating environment. 

Rethink preventive maintenance, starting with criticality 

A plan drawn up at commissioning, based solely on manufacturers’ recommendations, treats every asset the same way. The most effective way to review it is to start from criticality. 

The first step is to rank assets on two criteria: the consequences of a failure (safety, environment, production, cost) and its likelihood. For the most critical assets, apply reliability-centred maintenance (RCM), whose criteria are set out in SAE JA1011. This means identifying the failure modes and choosing the most appropriate task for each one: inspection, scheduled replacement or condition monitoring. For low-criticality, redundant assets, run-to-failure is often a perfectly rational choice, provided it is a deliberate and documented decision. 

Condition-based maintenance rests on a simple concept, the P-F interval. This is the time between detecting a potential failure (abnormal vibration, a rise in temperature) and functional failure, the point at which the asset can no longer do its job. If that interval is measured in weeks, periodic readings may be enough. If it is measured in hours, you need continuous monitoring, with alarm thresholds set by people who know the equipment. 

Get your data ready before rolling out AI 

A generative AI tool is only as good as the documents and history it draws on. Before rolling one out, four conditions need to be in place: 

For safety-related procedures, any answer the AI gives must cite its source and be signed off by an authorised person. 

These four levers share the same foundation: a reliable asset register and well-organised information. That is where to start, before choosing a tool. 

Get the foundations right before you tool up 

That is AMEXIO’s conviction. We have spent more than twenty years helping organisations manage and structure their information, and we believe maintenance performance depends first and foremost on the quality of your asset register. That is why we start by taking an inventory of your assets and reviewing the business rules already in place. We then work with you to build a data model that separates locations from assets, harmonises classifications and gives your technicians failure codes they will actually use. Your maintenance documentation is structured and linked to the assets it covers, so that everyone gets the right information at the right time. Tools and predictive analytics come next, on solid foundations. 

IBM Maximo, the platform to scale up 

To support this approach, AMEXIO, an IBM Gold Partner, relies on IBM Maximo Application Suite, which covers the entire lifecycle of your assets. Each of the four levers described above has a matching tool. 

Centralising knowledge with Maximo Manage and Maximo Mobile 

In Maximo Manage, work order history, sensor readings and documentation all come together in the asset record. Since version 9.1, Maximo Assistant has let you query that record in natural language, and the similar work order search brings up comparable jobs already carried out, along with their diagnosis. Maximo Mobile puts this information in the hands of technicians in the field. 

Improving failure code quality in Maximo Manage 

Maximo natively separates locations from assets and keeps an asset’s history when it is moved. Problem, cause and remedy codes are organised by asset class, and the AI service can recommend a code based on the technician’s description. 

Managing preventive maintenance with Maximo Health, Monitor and Predict 

Maximo Health calculates health, criticality and risk scores for the assets in scope, depending on the data available and the calculation rules chosen. Maximo Monitor tracks sensor readings. When those readings feed meters in Maximo Manage, a threshold breach can generate a work order, according to the rules configured. Maximo Predict estimates the probability of failure. Maximo Reliability Strategies helps you build RCM-based plans. Within it, AI can put forward an initial analysis of failure modes and their effects, which reliability teams then validate and complete with a criticality assessment. Finally, Maximo Visual Inspection analyses images from cameras, drones or mobile devices to spot cracks, corrosion or wear. 

Maximo Assistant in the field with version 9.2 

Maximo 9.2 has been available since 25 June 2026. With this release, technicians can use Maximo Assistant on their mobile device to search for information about an asset in natural language and review its maintenance history. The quality of the answers depends directly on the quality of the asset register and the history behind it, which is why it pays to get them in order first. 

Moving to IBM Maximo: where to start 

For an organisation moving from another CMMS, spreadsheets or in-house tools, the project rarely starts with the software. The first step is to migrate the existing master data: asset list, locations, job plans, spare parts inventory. This is the moment to apply the rules described above: remove duplicates, harmonise classifications and set up failure codes by asset class. 

Next comes the question of history. Migrating everything makes the project heavier; migrating nothing deprives Maximo Health and Maximo Predict of the data they need. What to migrate depends on the indicators and models you are aiming for. A good compromise is often to prioritise the history of the most critical assets. 

Finally, the rollout can be phased: a first scope, such as one site or one asset family, with Maximo Manage and Maximo Mobile, then extension to other sites and activation of the analytics modules once the data has stabilised. 

Licensing and deployment: AppPoints, SaaS or customer-managed software 

The licensing model makes this phased approach easier. AppPoints support a gradual rollout: you can start with work management, then extend usage in line with your needs, your entitlements and the number of AppPoints required. The suite is available as SaaS or as customer-managed software, and its APIs make it easier to integrate with ERP, SCADA or GIS systems. The approach works just as well for a smaller organisation as for a complex site such as an airport, with its many interdependent assets. 

Ready to move from reactive to predictive maintenance?

Join us in Luxembourg on Thursday 26 November. 
AMEXIO and IBM are hosting a lunch round table on AI in IBM Maximo, with real-world use cases. Places are limited, so register now.