Every one of them was correct when it was issued.
Save the customer at all costs. Then: no more free work, we're bleeding margin. Then: avoid refunds. Then: avoid disputes, take the refund if you have to. Then: restart billing or offboard them.
Five mandates, six months. Each one issued by someone reasonable, responding to real information, in a room where the previous mandate's consequences had just landed on the P&L. Nobody was being erratic. Every reversal was a correction.
And the accounts don't reset between them.
What's Actually Sitting In Your Book
The customer who signed in March got two months free under save-at-all-costs. The one who signed in April got a partial refund because we were avoiding disputes that week. The one from May got their start date pushed and two payments banked forward. June's cohort got nothing, because by then we weren't making concessions.
Same product. Same price on the order form. Four different deals.
Now try to answer a basic question about that book. What's the actual ARR? What does renewal look like for the March cohort? Which of these accounts owes you money and which of them believes, with documentation, that they don't?
Nobody can answer. Not because the records are bad but because the terms were never written down as terms. They were written down as exceptions, one at a time, under five different theories of what mattered.
The Part That Gets Missed
The finance cost is the small one. The expensive part is that your customers learned something.
A customer who received a concession under one mandate now knows concessions are available. They don't know the mandate changed. So when they ask again under the new regime and get refused, they haven't experienced a policy update. They've experienced you going back on something.
That's how a company that spent six months trying to save customers ends up with a book that trusts it less than when it started. Every individual decision was made to preserve the relationship. The sequence taught the customer that the relationship has no rules.
And it lands hardest on the accounts you spent the most on. The ones who got the most flexible treatment were, by definition, the ones you fought hardest to keep.
Where It Ends Up
Run this long enough and something quiet happens to the book.
Every customer who stayed did so under some arrangement. A credit, a pause, a start date that moved, a rate that was never the rate. Meanwhile the ones paying full price on standard terms had no reason to escalate, so nobody ever built them a reason to stay, and a good number of them left on schedule.
What's left is a retained book and a paying book that are no longer the same book. The logo count looks survivable. The revenue doesn't. And nobody can point at the month it happened, because it didn't happen in a month. It happened one exception at a time.
Why It Happens
Mandate churn is a symptom of a company treating customer-facing policy as a lever for a number that's due this quarter.
When cash is tight, concessions look expensive, so they stop. When churn spikes, concessions look cheap, so they resume. When disputes hit the merchant account, the calculus flips again. Each of those is a rational response to the metric that's screaming loudest, and none of them are decisions about what kind of company you are.
The tell is easy to check: your policy changes track your internal metrics rather than anything about your customers. If the terms a customer receives depend on which month they escalated in, the policy isn't a policy. It's a running average of your worst weeks.
What Should Have Been Stable
Almost all of it. The specific concessions can vary by situation. The rules governing them shouldn't.
What triggers a make-good. What a make-good can consist of. Who can authorize one and up to what value. What happens when you fail to deliver something you sold, not case by case, but as a standing answer. Whether a paused account keeps its original terms.
Those are decisions about how the business behaves. They can be revised deliberately, announced, and applied forward. What they can't survive is being re-derived monthly by whoever's under the most pressure.
What Stable Terms Actually Buy You
Not the absence of the above. Something you can only do once the rules hold.
A customer who knows the rules can be sold to. That's the whole return on it. Expansion asks somebody to believe that what you tell them today will still be true in six months, and every reversal is a live demonstration that it might not be. You can have the best next thing in your category and it won't move, because the person you're asking has already learned what your commitments are worth.
Asking a customer for more doesn't damage trust. Being unpredictable does. Which means stable terms aren't a defensive measure or a tidiness exercise. They're the precondition for asking anyone for anything, the same way delivery has to be real before an expansion motion built on top of it survives contact.
Go Look
Pull every account that received a non-standard arrangement in the last twelve months: refund, credit, pause, discount, extension, make-good.
For each one, write down what they got and what was driving the decision that month.
Then group them by month. If the pattern in that column changes three or four times across the year, you don't have a policy. You have a record of what your leadership was worried about, written into your customer relationships, one account at a time.
Lincoln Murphy formally named and popularized Customer Success starting in 2010 and has spent 15 years connecting it to expansion revenue and commercial outcomes. Read The Premise.