Once brought in, Managed Service Providers (MSPs) are very difficult to remove. One of the reasons for this is they have tons of metrics to show upper management how much value they bring. Charts and graphs presented to management tout the number of tickets opened and closed, average response time, aggregate server up time, and other details derived from the ticketing system itself, and finally total MSP cost versus that of dedicated staff.
What is usually not tracked is user downtime, and the cascade effect it has on the organization as it fans out to other users and departments.
Rather than comparing dedicated-staff costs vs MSP costs, we should be comparing aggregate corporate efficiency under dedicated-staff support vs MSP support.
But again, that's hard because it isn't tracked. And therein lies the problem.