From my vantage point (selfishly no doubt since my team has created it), the Qlik Sense Telemetry Dashboard (https://github.com/eapowertools/qs-telemetry-dashboard/) is the best tool for the job. Now the resulting information will not be in percentage terms, mostly because that isn't a super helpful measure by my estimation. The ultimate care isn't what precise % of available compute is used, it's how long do users have to wait.If the system could be pegged at 100% for the rest of time and every user got a response time of under 2 seconds, then that would not be a bad state of affairs. For a primer on what the dashboard provides, do review this video: https://youtu.be/2BfZDxODhzE
There are other methods for measuring idealized response times for a Qlik app (Qlik Scalability Tools, Rob Wunderlich's Document Analyzer), but those are providing a slightly different data set. They focus on, in an idealized scenario, what is the response time of an object. When doing regression testing or performance tuning, it's ideal to have this isolation so that you can have a baseline between trial X and trial X+1.
But for day-to-day awareness of what objects (think visualizations) and apps are the most computationally expensive, then the Telemetry Dashboard are my preferred tool.
