In some cases the customer could have implemented daily performance analytics collection jobs that show to run just a few minutes at the start of the month but by the end of the moth take hours if not days to complete.

This is likely happening when a daily job is configured based on monthly indicator sources with a data filter like "created on this month".
When using this for daily collection the data set for the collection to process is increasing from the first to the last day of the month and this could cause the worker job to run over 24 hours conflicting with a daily job cycle as well as causing alerts to trigger for service now staff to investigate (on a monthly basis), not to mention unnecessarily draining instance performance.

To solve this the solution is quite simple, just ensure that for daily data collection you are using an daily indicator source rather than one that is there for monthly collection.