In this guide 6 sections
A plug-in sensor with its own app is enough when one person watches a few readings and nothing else depends on them. Choose custom when alarms must follow the shift roster, readings must come from a PLC, or sensor data must raise a work order in your ERP. Check the exit cost first.
Where a ready sensor with an app is enough
A Wi-Fi temperature tag, a clamp-on energy meter or a tank level sensor can be live in an afternoon with no engineer on site. The app draws the trend. Alerts reach one phone. For a single cold room with one owner and a low penalty for a missed reading, that is usually all the monitoring you will ever need.
It also suits a trial. A small kit settles whether readings change any decision, and it does so before anyone writes a specification or books a site visit.
The test is short. Name the one person who watches the screen and the exact action they take when a high-temperature alarm sounds at two in the morning. Clear answers mean a kit will do.
- One site, a handful of points
- One viewer, no shift roster
- No link to other software needed
- Loss of a day’s data is tolerable
Signs you have outgrown the plug-in app
The first sign is people. When three shifts need different alerts, a vendor app that sends everything to one phone stops working, and you need routing by roster with escalation to a supervisor. Someone must record who acknowledged the call.
The second is machines. If readings must come from a PLC, a VFD or a controller over Modbus, a clip-on tag cannot reach them and someone must write the register map. That is engineering work.
The third is the site. A plant with a poor network needs an edge gateway that decides locally by starting a standby pump, and syncs to the server once the link returns. The fourth is audit pressure. When a customer or inspector asks for a calibration record, a screenshot from a phone app will not satisfy them, whatever the graph looks like.
Integration with the rest of the business
Ready apps keep data in the vendor’s cloud and format, which is fine until bearing vibration should raise a maintenance work order or a batch record should carry humidity. Those links need an interface to your ERP, maintenance system or quality records, so ask whether the vendor offers a REST API, a CSV export or nothing at all.
Test one real export. Open the file and read it.
Custom work earns its place only when sensor data has to trigger something in another system, such as a work order, a stock hold or a batch release. Data that is only viewed rarely needs it.
- Work orders raised from alarms
- Batch or lot records with conditions attached
- Stock or dispatch holds on a failed chamber
- Reports in your own format
What it costs to switch later
Starting small is not wrong. The real question is what you lose when you outgrow the kit, because hardware often locks you to one cloud and the export may carry only hourly summaries.
Ask three things before buying. Can you download raw readings, do the devices speak a standard protocol such as MQTT or Modbus, and what happens to the data if the subscription lapses?
If the answers are poor, treat the kit as a trial, plan to start the history again later and build no procedures around it. Check who owns the firmware too. A kit that stops receiving updates becomes a security question, not only a cost question, once it sits on your plant network.
A decision you can make this week
List the points you want to monitor, who acts on each alert, which systems must receive the data and how long you need the history. Simple answers mean buy the kit. Run it for a month.
If two or more answers are complex, write them down as requirements, get a custom design quoted against them and compare both quotes line by line.
Your next step is the comparison sheet. Fill it in for the first five points you care about and mark every row where the ready option fails, because those rows are your custom requirements.
Ready or custom comparison sheet
Print one row per monitored point. Columns cover who acts on an alert, shift routing, source device and protocol, receiving systems, outage behaviour, history length and export format. A last row asks about vendor exit. Complex rows point to custom work.
Open a blank worksheet to printQuestions owners ask
Can we start with a ready sensor and move to custom later?
Yes, provided the devices use standard protocols and you can export raw readings with time stamps, otherwise expect to restart the history and rewire or replace some hardware.
Do ready sensors work without internet?
Many need the vendor’s cloud to show graphs or send alerts, so ask what the device does during an outage and whether it stores readings locally.
When does an ERP link justify custom work?
When sensor data must trigger an action elsewhere, such as a maintenance order, a stock hold or a batch record, an ERP link justifies custom work. Viewing alone rarely does.
What should I ask a sensor vendor before buying?
Ask about raw data export, open protocols, alert routing by shift, offline behaviour, support length after sale and what happens to your history if you cancel.