Almost every exit conversation assumes a sale. There is another version nobody writes about, where you are done and there is no buyer worth the trouble. Twelve clients, revenue that walks out the door with you because you are the technical department, and a stack somebody else would have to unwind before they could bill against it. Or a buyer does exist and the offer comes with a two-year earnout you would rather not sign.
That business is often worth more wound down properly than sold badly. It is worth close to nothing if you just stop answering the phone, which is what a lot of these turn into.
I should say up front that I have not closed an MSP. I have bought a few and taken clients on mid-transition, and the gap between a clean handover and a bad one shows up in the first week. Most of what follows is the version I wish had been true on the receiving end.
Pick the date before you tell anyone
Not a season, a date. Everything below gets scheduled backward from it, and without one you will drift for a year while getting steadily more tired.
Ninety days is the floor and a hundred and twenty is realistic. Under sixty and you are going to hand somebody a mess, and they will talk about it.
Read your own paper first
Before you talk to a single client, go read what you actually signed.
Your client agreements: notice period for termination, auto-renew language, and whether you can assign the contract to another provider or whether the client has to sign fresh with them. Most MSP agreements are month to month with thirty days, but the two that are not will turn out to be your two biggest clients.
Your vendor agreements matter more and get skipped. Microsoft NCE annual terms do not care that you are closing, and neither does your RMM contract or the three-year deal you signed for the PSA. Write down every commitment that outlives your end date and what it costs to carry or break it. That number is part of the decision.
Inventory what you are holding that is not yours
This is the actual work of not stranding people, and it should happen before any announcements:
- Domain registrars and DNS, especially anything sitting in your account instead of theirs
- Microsoft partner relationships, GDAP, and which subscriptions bill through your tenant
- Admin credentials, break-glass accounts, and anything with your email as the recovery address
- Backup repositories, retention dates, and where the data physically sits
- Documentation in your PSA, Hudu, or IT Glue
- Firewall and switch management accounts, plus ISP and phone accounts opened in your name
- Any hardware on their site that you own or are still financing
- Certificates and anything else that expires and would quietly break in four months
Every line on that list is a way to strand somebody after you are gone.
Find landing spots before you say a word
Plural. Do not hand thirty clients to one MSP unless that MSP can genuinely absorb thirty clients, because they will take the deal and then serve the ones they like. Two to four receiving providers is normal. Match on size, vertical, and stack, so your one dental office goes somewhere that already runs dental.
Talk to the owners directly and tell them what is wrong with each account, including the difficult ones. They will find out anyway, and the thing you cannot afford is a provider who feels lied to and drops your former client six months in.
Get the referral fee in writing before any introductions
A wind-down can still pay. Not sale money, but real money for a book nobody could have bought.
What I have seen work is one to three times the monthly recurring per client that signs and stays, paid out over six to twelve months and tied to retention so nobody takes the logo and dumps the account. A percentage of first-year revenue works too. Whatever the structure, it is worth nothing as a handshake, because you will have no leverage at all the day after the transition. Sort it before the first introduction. After that, you are asking for a favor.
Tell people in the right order, on the same day
Staff first, in the morning, in person. They will find out anyway and they need to know what happens to them, including whether the receiving firms are hiring.
Then your top clients by revenue and relationship, by phone or in person, before anything is in writing. Those calls are short: I am closing on this date, here is who I recommend for you and why, and I will hand over everything either way.
Then everybody else in writing, the same day. Do not stagger it over two weeks. Clients talk to each other, and finding out secondhand turns a professional exit into a rumor about you.
What the letter says
Five things. The date you stop supporting them. A one-sentence reason with no drama in it. Who you recommend and why that provider fits them specifically. What happens if they choose somebody else. And that you will hand over every credential, document, and piece of their data regardless of who they pick.
That last one is the part people quietly skip when a client picks a provider they do not like, and it is the part everyone remembers.
The handover
Credentials go into a shared vault, not a spreadsheet and not a PDF. Rotate anything you personally knew once they have it, or tell them to.
Documentation needs to be exported in a form the next provider can actually read. A PSA export nobody can import is not documentation.
Microsoft licensing is the one that bites. There is a partner-to-partner transfer path in Partner Center, but it does not cleanly cover every SKU or every term, and when it does not work the practical answer is that the new partner provisions fresh subscriptions while your annual commitment keeps billing on your side until the term runs out. Find out which situation you are in per client early, because the answer can move your end date.
Backups: either restore their data somewhere they control, or tell them in writing the exact date your retention drops it. Do not let a repository quietly expire in month four.
Anything in your name goes into theirs, including registrar, ISP, phone, and cert renewals. This part is tedious and it is most of the last two weeks.
Set a real end date and a short tail
Support ends on the date. Then offer a short emergency-only window, two or three weeks at an hourly rate, so a genuine outage during cutover does not go unanswered.
Without that boundary you will still be doing free work in March for the two clients who did not move, because they will not move until they have to.
The parts that get missed
Collect your receivables before the end date. Leverage disappears the moment you stop being their provider.
Cancel your own subscriptions on a schedule rather than all at once. You need the RMM, the documentation system, and ticketing alive until the last client is fully off, and killing the docs platform to save a month of fees is how you end up unable to answer a handover question.
Buy tail coverage on your E&O policy and do not dissolve the entity the day you finish. Claims can arrive after you close, and a dead LLC with no policy behind it means it lands on you personally. Your broker and your accountant should tell you how long to keep both alive.
Decide what data you keep and for how long, tell clients, then actually delete it on that date.
The test
On day one without you, can each client get into their own systems, does anyone know how their environment is built, is anything critical still billing to your name, and do they have somebody to call.
If all four are fine, you closed well. This industry is small enough that people remember either way.
Has anyone here actually gone through this? I am curious what broke that you did not see coming, especially around licensing.