Condition monitoring: what to check before buying industrial sensors
Before investing in sensors or predictive maintenance software, define the equipment problem, the information you need and who will act on an alert.
Table of contents ExpandCollapse
Before investing in sensors or predictive maintenance software, define the equipment problem, the information you need and who will act on an alert.
Begin with the maintenance question
A sensor can produce more readings without making a maintenance decision easier. Start by asking what the team needs to know: whether a particular asset is changing, whether a recurring problem can be investigated earlier, or whether inspections are reaching the right equipment. Choose an asset with a documented operating history and a clear maintenance owner. Avoid starting with a large deployment simply because a supplier offers a bundle.
Separate observation from prediction
Condition monitoring follows equipment behaviour; diagnosis investigates what an observed change may mean. Predicting future condition is another task with its own evidence requirements. NIST researches these capabilities as related but distinct areas. A dashboard showing a trend should not automatically be treated as a reliable forecast of failure. Ask the vendor to explain exactly which capability is being supplied and how it has been evaluated.
Review the records already available
Collect service notes, inspection findings and existing machine data before choosing additional measurements. Check whether timestamps, equipment identifiers and operating states are recorded consistently. Changes in load, speed or production conditions can affect interpretation. A qualified maintenance specialist can help decide whether a proposed measurement is relevant to the failure mode being investigated, rather than merely easy to collect.
Specify the response to an alert
An alert is useful only when somebody knows what to do next. Write down who receives it, who investigates and how the finding is recorded. Decide what happens outside staffed hours and how repeated unhelpful alerts will be reviewed. This is a maintenance decision process, not an instruction to override existing controls. Monitoring should complement approved safety procedures and manufacturer maintenance requirements.
Check installation and information handling
Have qualified personnel review mounting, environmental suitability, power, access and any connection to existing systems. Ask how measurements reach the application, what happens if connectivity is lost and whether local records remain available. Confirm update responsibilities and access permissions. Do not attach equipment, open guards or alter machine controls as part of an informal trial; installation belongs within the site’s approved engineering and safety process.
Evaluate a bounded pilot against a baseline
NIST’s maintenance integration white paper describes the value of defining use cases and establishing current capabilities before deployment. For a pilot, agree what evidence would support continuing: better investigation records, useful lead time for an identified issue or clearer inspection priorities. Keep a record of missed events and false alerts as well as apparent successes. A short trial with no equipment problems cannot demonstrate that every failure would have been detected.
Compare the ongoing commitment
Check calibration or verification needs, replacement parts, subscriptions, support and the effort required to interpret the data. Ask for a readable export so that your maintenance history remains usable if you change providers. Review the pilot with the technicians who acted on its findings. This article is a research-led buying framework, not an engineering specification, a safety assessment or a claim that predictive maintenance eliminates breakdowns.
Explore Industry Technology for related context, or read our guide to a first digital manufacturing pilot.
