What an MSP Scope of Work Should Actually Say
Every scope dispute in a managed services relationship traces to a task both sides assumed the other owned. An MSP scope of work earns its keep on the exclusions page, not the inclusions list.

Ninety days is a realistic plan to offboard a managed service provider. Compress it and the revocation step is what gets cut, leaving agents installed and credentials live for months.
To offboard a managed service provider, work in this order: read the contract, inventory the access, take ownership of the accounts, extract the documentation, then move the services. Most failed transitions go in the opposite order, starting with migration and discovering at the end that credentials, domains, or licenses were never yours.
I am Amelia S. Gagne, CEO of Kief Studio. This is the operational sequence. The contract language that makes it possible is technology vendor contract exit terms, and the selection-stage version is how to evaluate a managed service provider.
Before notifying anyone, know what the contract says about notice period, termination fees, transition assistance, data return, and what survives termination.
Two details decide the shape of the whole project. Whether transition assistance exists and at what rate, and whether it survives the specific type of termination you are exercising. Assistance that applies only to termination for convenience is a different document from assistance that applies to termination for cause.
The New York Department of Financial Services issued guidance on third-party service provider risk on October 21, 2025 directing covered entities to review the provider agreement before termination to identify offboarding obligations and protections, and to develop a transition plan with defined timelines and roles. It binds regulated financial entities and describes a sequence any operator benefits from following.
If transition assistance is absent, you are asking for cooperation rather than invoking it, and that changes both the timeline and the tone. Plan for a longer window.
Do this before notification, not after. Not because providers behave badly, but because the inventory is far easier to build while the relationship is still routine.
What to enumerate:
The recurring finding is licensing bundled under a provider agreement, where leaving means repurchasing rather than transferring. Knowing the number before you negotiate the exit is better than learning it during. The ownership principle is in who owns the site, the host, and the login.
Ownership transfers first, migration second. An account you do not control is a dependency that outlives the contract.
Move the registrar account, the identity provider tenant, and the cloud billing relationship into accounts controlled by your organization, with recovery addresses on a domain you own. Use a role-based address rather than an individual's, so this does not become a personnel dependency later.
Where a service genuinely cannot transfer, that is a migration item rather than an ownership item, and it should move to the project plan with a cost attached rather than being noted as unresolved.
Ask for it in a portable format and check what arrives against what you asked for.
The useful list: network diagrams, firewall and switch configurations, an application and integration map, backup job definitions and schedules, monitoring and alert configurations, the asset inventory, open ticket history, and the credentials inventory.
Knowledge transfer sessions are worth scheduling while the outgoing provider is still contractually engaged, and worth recording. The question to ask in them is not how the system works, which is documented, but what is unusual about this environment. Every environment has three or four undocumented quirks, and those are what break a migration.
Where the contract allows, a direct conversation between the outgoing and incoming providers saves weeks. Not every outgoing provider will agree, which is why the clause matters. This is the same asset problem as institutional memory that survives turnover.
Run the new arrangement in parallel where you can afford to. Overlap costs a month of double fees and removes most of the risk from the cutover, which is a good trade for anything the business depends on. It is the reversible version of the decision described in two-way doors.
After cutover, revoke deliberately rather than by attrition:
Then get written confirmation of data deletion. This is the step everyone skips and it is a live liability. Data left on a former provider's systems remains your notification problem if it is breached, and the relationship ending does not end the record. That is the cheapest way to protect data is to not keep it, applied to storage you do not control.
For a small or mid-sized organization with a single provider, ninety days is a realistic plan: two weeks on contract review and inventory, four to six weeks on ownership transfer and documentation, four weeks of parallel running, then revocation and confirmation.
Compress it and the failure mode is predictable. The revocation step gets abbreviated, agents stay installed, credentials stay valid, and the exposure persists quietly for months.
Tell your cyber insurance carrier if the transition materially changes who operates your controls, since that is the kind of change that belongs in a mid-term notification rather than a surprise at claim time. The reasoning is in what a renewal application asks now.
A good provider makes leaving straightforward, because a relationship held together by switching cost rather than by results is weak for both parties. We build for that on our side, which is part of why the operating model at ltfi.ai keeps client data on infrastructure with a clear export path. The reasoning is in what owning the stack means for client data.
Brian Gagne handles the infrastructure and identity side of transitions and has for the fourteen years we have worked together, and briansgagne.com is the deeper technical read. Developer tooling we publish openly is at kief.dev.
In sequence: review the contract for notice, fees, transition assistance and data return; inventory all provider-held access before giving notice; transfer account ownership; extract documentation and run knowledge transfer while they are still engaged; then migrate, revoke access, and obtain written deletion confirmation.
Registrar and DNS, identity provider administrative accounts including break-glass, cloud tenant and billing ownership, licensing held under the provider's agreement, backup location, monitoring and remote access agents, and any certificates, service accounts, or API keys issued under their control.
Around ninety days for a small or mid-sized organization: two weeks for contract review and inventory, four to six weeks for ownership transfer and documentation, four weeks of parallel running, then revocation and confirmation. Compressing it usually shortens the revocation step, which is the one that leaves exposure behind.
Licensing purchased under the provider's agreement rather than yours, remote monitoring agents left installed on endpoints, credentials that are disabled but not rotated, and the absence of written confirmation that your data was deleted from their systems.
Because breach notification obligations follow the data rather than the contract. Records left on a former provider's systems remain your exposure if they are compromised, so written deletion confirmation belongs in both the agreement and the offboarding checklist.
Every scope dispute in a managed services relationship traces to a task both sides assumed the other owned. An MSP scope of work earns its keep on the exclusions page, not the inclusions list.
Six tasks get dropped in nearly every co-managed IT arrangement, starting with deprovisioning on someone's last day. Assigning them explicitly takes an hour and closes most of the seam.
NetDiligence found business interruption incidents cost small and mid-sized organizations over five times more than incidents without it. Duration is the expensive variable, and an MSP SLA usually does not commit to it.
Work With Us
Kief Studio builds, protects, automates, and supports full-stack systems for businesses up to $50M ARR.
Newsletter
Strategy, psychology, AI adoption, and the patterns that actually compound. No spam, easy to leave.
Subscribe