There is a specific moment that brings most organisations to question their MSP relationship. It is not an outage. It is the monthly service review where every metric on the dashboard is green — response times met, tickets closed, availability delivered — and somebody senior finally says what everyone in the room is thinking: then why is everybody complaining?
IT service management has a name for this. The watermelon effect: green on the outside, red on the inside. The service level agreement is technically being met while the actual experience of the business deteriorates. It is one of the most widely recognised failure modes in outsourced IT — written about across the ITSM industry for years — and it persists because nothing about it requires bad faith.
The metrics are accurate. They are measuring the wrong thing.
Consider a standard first-response SLA. A user's laptop won't start. A ticket is raised, acknowledged within eight minutes — comfortably inside target — and closed the same day. Green tick. The laptop still doesn't work; the user raises another ticket; that one is also acknowledged promptly. Two green data points, zero working laptops.
Multiply that pattern across a contract and you get a reporting pack that is simultaneously true and misleading. Response measured instead of resolution. Tickets closed and reopened rather than fixed. Incidents reclassified into categories with softer targets. The clock stopped while a ticket sits "awaiting customer". None of this necessarily involves anyone gaming anything deliberately — target-driven behaviour bends toward the target, and SLAs that measure activity rather than outcomes reward activity rather than outcomes.
There is a second, structural problem sitting underneath: in most MSP relationships, the party being measured is also the party doing the measuring. The provider runs the toolset, defines the categories, generates the report and presents it. That is not an accusation — it is simply an arrangement no finance director would accept anywhere else in the business.
What it costs
A green dashboard is not a neutral decoration. It is the evidence base for real decisions: whether to renew, whether to escalate, whether the complaints from the sales floor are a service problem or a perception problem. When the measurement is broken, those decisions are made on false signal. Renewals get signed on the strength of reporting packs. Underperformance gets normalised because it never appears anywhere a board member would see it.
The industry's answer has been a gradual shift toward experience-level agreements — measuring how the service actually feels to the people using it, alongside the contractual mechanics. That direction is right. But it is a second step. The first is establishing whether the numbers you already receive mean what they appear to mean.
Testing measurement integrity
This does not require a transformation programme. It requires sampling. Take a month of closed tickets and walk a proportion of them end-to-end: was the issue actually resolved at first close? How many reopened under a new reference? How were incidents categorised, and did the categorisation match the impact? Compare the reported user experience with a straw poll of the users. Ask for raw ticket data, not the summary pack, and see whether the request lands as routine or as an affront.
The pattern of answers tells you most of what you need to know. A provider confident in its own numbers has no difficulty showing the workings. Measurement integrity — whether the reported performance reflects the experience of the business, and whether the measurement itself can be trusted — is one of the specific things an independent inspection tests, because it sits underneath everything else. If the numbers can't be trusted, neither can anything built on them.