The three questions every MSP contract should answer clearly
If your MSP's contract doesn't answer these three questions inside two minutes of reading, it's the contract - not the service - that's going to hurt you.
We read a lot of MSP contracts. Most of it is recycled boilerplate, some of it is quietly predatory, and almost all of it skips three specific questions that decide what happens when the relationship goes sideways 18 months in.
Here are the three. Give yourself two minutes with any MSP contract and see if you can find clean answers. If you can't, you've already learned something useful about the provider.
1. What does the exit clause actually cost?
Most MSP contracts have a 30-day termination clause. That bit is fine, and usually the bit prospects focus on. The trap is the next paragraph: “data extraction, credential handover, and disengagement assistance are quoted as a professional services engagement at our standard hourly rates.” That sentence translates to between five and fifteen thousand dollars of work just to leave.
Cost is only half the problem. The format the data comes back in matters more. We've seen offboarding packs delivered as PST mailbox exports, scanned PDFs of network diagrams, and credential lists pasted into an email. Technically a handover. Practically useless to the next MSP.
A good exit clause names what you get back, the format, the timeline, and the cost. What you get back: admin credentials, asset register with serials, network diagrams in editable form, M365 and Azure tenant role reassignments, third-party licence transfers, ticket history exports, configuration baselines. Format: machine-readable - CSV, JSON, .docx, draw.io or Visio for diagrams. Timeline: 14 to 30 business days from termination. Cost: explicitly at no additional charge, or capped at a named fixed fee.
2. Who owns the operational evidence?
This is the question almost no contract answers, and it's the one that decides whether your next MSP transition is two weeks or two months.
From the day an MSP starts work, they're building an operational moat: every config they apply, every change they push, every alert that fires, every ticket that gets closed, every runbook they write. Eighteen months in, that moat is the thing that makes switching slow, expensive, and risky. The next MSP starts blind because the old MSP is still holding the maps.
The contract should be explicit that the operational evidence created in delivering the service is your property, not theirs. Configuration baselines, ticket and alert history, change logs, monitoring dashboards, runbooks. You should have read access to the MSP's PSA and RMM tools as a matter of course - you shouldn't need to ask to see your own ticket history. And on exit, all of it should export in formats the next provider can actually load.
3. What's the response SLA, and what's the SLO behind it?
Most MSP contracts publish an SLA - the contractual commitment - and stop there. The SLA is the floor: the slowest response the MSP can give before they breach the agreement. What's missing is the SLO: the operational target the team actually works to.
The trap is subtle. If the P2 SLA is “respond within four hours”, an MSP can hit it 100% of the time at three hours and 59 minutes and never technically breach. The number on the contract looks great. The lived experience is awful.
A good contract publishes both. SLA per priority: P1 = business-critical outage, P2 = major impact on multiple users, P3 = single-user impact, P4 = scheduled work, with a maximum first-response time against each. SLO per priority: the internal target the operations team works to, typically tighter than the SLA by a factor of three to ten. Plus an explicit measurement methodology: “first response = human acknowledgement via ticket or phone, not automated receipt.” Bonus points if the MSP publishes a rolling 90-day report of actual performance against both numbers.
How to use this
Two minutes per question. If you can't find clean answers in the document, ask the MSP directly. How they answer tells you more than the contract does. If the verbal answers are good but the document is vague, insist on a side-letter that commits them to the verbal answer in writing.
Our own contract answers all three in the first three pages. It's not magic, and it's not generous. It's just the right set of things to be specific about, written by someone who has read enough of the other kind to know what gets ducked.