Almost any machine can be instrumented. The question is whether the data it produces will justify the work, and that is a question about the process around the machine rather than the machine itself.
What decision will the data change?
If nobody can name the decision, the project has no benefit case. "Visibility" is not a decision. "Whether we run the second shift on line three" is.
Who owns the number once it exists?
Data with no owner degrades. Within a year the sensor has drifted, nobody has noticed, and the dashboard is quietly wrong. Name the owner before the sensor is installed.
What is the machine's actual failure history?
Instrumenting for a failure mode that has never occurred is common and expensive. Ask maintenance for three years of work orders before choosing what to monitor.
Can the machine be taken out of service to fit it?
The technical design is often the easy part. The scheduling is not. If the only available window is the annual shutdown, that determines the programme timeline.
What happens when the connection fails?
Adding a network path to a machine adds a failure mode. The design must specify how the machine behaves when that path is unavailable, and the answer must be acceptable to operations before the work is approved.
None of these questions require an engineer to answer. They do require somebody to ask them early, and in our experience the projects that go well are the ones where they were asked before the budget was set rather than after.