Every business I work with has an onboarding process. Someone has thought hard about the kick-off call, the welcome pack, the first ninety days. Almost none of them have thought about the last ninety.
The exit is the only part of a commercial relationship that reliably becomes public. Everything else the client experienced happened in private: the good months, the problem you fixed at eleven at night, the invoice that was wrong and got corrected without fuss. Nobody repeats those at a conference. They repeat how it ended.
Ending well is a different thing from ending softly. Clear is kind. Nice is often not.
I have been on both sides of this.
The customer side
The instinct on a bad exit is scorched earth. Press release, public campaign, vendor as villain. I drafted mine on a Tuesday. It read well. It also asserted about six things I could not have defended under a single question from anyone who practises law for a living. I had to be talked down, and I am glad I was.
What works instead is evidentiary restraint. A dated chronology. Correct statutory references. No adjectives about the product, no speculation about motive, no rhetoric. You read the contract closely enough to know which parts of it actually bind you, and you negotiate from there. That is not a departure from the rules. It is using them as they stand, which almost nobody bothers to do, because reading the contract is dull and being angry is not.
Two traps, both of which I walked into.
The first is that the arguments which feel strongest are usually the ones a lawyer identifies as invented. The two that felt like knockout blows to me were the idea that a vendor with no visible costs cannot charge for its trouble, and the idea that a fee I resented was therefore a penalty. The strength was manufactured by the grievance. The file decides what is strong.
The second is that your own past correspondence is the counterparty’s best evidence. Every reasonable email you ever sent belongs to them now. Every “no problem, happy to work around that” is an exhibit. The case has to be built out of what survives contact with the file, and what you feel is not in the file.
The vendor side
This is the more useful half, because more of you are standing on it.
The failure in a bad exit is rarely greed and rarely laziness. It is structural. Take a company big enough to have a billing queue. The person working that queue has a screen, a policy and a number to hit. Somewhere else, in another timezone, on a recording nobody has opened since the day it was made, a salesperson made a commitment about what would happen if this client ever wanted to leave.
Nobody decides not to check the recording. The process has no step where anyone checks the recording.
That is the whole failure. The commitment lives in sales, the exit lives in billing, and between them sits a gap that belongs to nobody. The relationship dies in that gap, politely, at the hands of someone applying the policy correctly.
Every scaling services business walks into this. It arrives at the moment the founder stops taking the calls. Before that, the person who made the promise and the person enforcing the invoice were the same person, and the system worked because it was one head. After it, the promise and the enforcement sit in different departments with different metrics, and nobody notices until a client leaves and tells the market why.
So: a service guarantee is worth nothing unless the person answering the phone can bind the company to it. If your front line has to escalate to honour your own marketing, you do not have a guarantee. You have a claim.
Why this matters more in strata
Strata is a small market with a long memory and committees that talk to each other. The exit process is the single most reputationally load-bearing thing a manager does, because a departing client is by definition talking to the market. They are running a tender. They are on the phone to three other managers. They are answering questions from a committee that has never met you and is forming its entire impression of you from one story.
The exit conversation is the only one your departing customer will repeat to other people.
That conversation happens whether you handle the exit carefully or not. The only thing you control is what goes in it.
Churn is changing shape
Churn used to be mostly about loyalty. Increasingly it is about capability. People leave because they want to try something else, including building it themselves, and that is a real category now rather than a threat somebody invented for a slide. Some of those attempts work. Plenty do not.
Which means a departing client is often a client on a trial run. A share of them will be back inside two years, and the only thing that decides whether they call you or one of your competitors is how you behaved on the way out. You cannot win them back with a campaign at that point. The decision was made in the handover.
You write the welcome pack knowing exactly who is going to read it. The last email you send gets read aloud in a room you will never enter.
Alex