Pick the approach that can reliably answer the operating question and that someone can support after installation. A packaged monitor can be the right choice even if you could build the same function. A gateway earns its place when existing equipment already produces useful data. Custom hardware or firmware is worth considering when a material requirement remains unmet.
Compare three options against the same job
- Buy a packaged monitor
- Choose this when suitable sensing, alerts and reporting already exist in a supported product. Check sensor placement, local operation during outages, who receives alerts, how you export history and what happens when a subscription ends.
- Add a gateway
- Choose this when a supported controller or meter already exposes the right readings but the rest of the business cannot use them. Check compatibility with the exact equipment, whether access can be restricted to reads, how records retain timestamps during outages and who maintains the mapping.
- Build a custom device or integration
- Consider this when the packaged options cannot meet a documented requirement, such as a specific signal combination, power constraint, installation environment or offline behavior. Include enclosure, wiring, commissioning, secure provisioning, updates, replacement and support in the design, not just the sensor and board.
Illustrative example: three assets, one useful decision
A fictional workshop wants its duty lead to know whether a compressor is running after closing. A packaged monitor can read a verified status signal and notify the lead, so it is the first option tested. A second asset already has a documented controller feed, so the workshop tests a gateway before adding another sensor. A third asset lacks a supported interface and sits where the proposed off-the-shelf monitor cannot operate as intended. Rather than committing to a custom build, the team first asks whether monitoring that third asset would change an action worth paying for.
The workshop evaluates each option with the same evidence: a witnessed state change appears correctly, lost connectivity is visible, and a named person receives an actionable message. Different assets may justify different approaches; using the same dashboard does not require the same hardware everywhere.
Compare the work after the purchase
List acquisition, installation, network access, recurring fees, calibration or sensor replacement, updates, data retention and support. Ask who can restore operation if a device is replaced, and whether you can retain records and configuration when changing providers. For custom work, name the owner for firmware, credentials and field replacements before approving a prototype.
Write down what makes the pilot worth continuing
Use one asset, one decision and a defined observation period. Agree how the reading will be checked, what an outage should look like, who responds and what evidence would justify expanding the installation. If an existing product meets that test and can be maintained, keep it. If it fails a material requirement, record exactly what failed before deciding whether to integrate, build, or stop.
Use the equipment monitoring guide to frame the pilot, and embedded systems for equipment monitoring to explore the requirements that might justify custom engineering.