Track the exception rate. Percent migrated tells you what the contract says. Exception rate tells you whether anyone believes it.
The formula is unglamorous: contested cases, off-plan discounts, manual credits, and side agreements, counted per week, divided by active accounts.
It works because an exception is the only artifact that gets created when someone does not believe the model. Nobody files a carve-out for a price they expect to hold. The exception is the belief gap, written down, timestamped, with a name on it. That makes trust countable on day one rather than something you assess by feel at the end.
What percent migrated hid
We killed a per-seat tier at a $40M ARR company, twenty-one months into a move to outcome pricing. Five things broke that were not in the plan:
- Twelve accounts had migrated on paper and were barely using the product.
- Fourteen of thirty-five mid-market accounts with edge use cases churned.
- The unit definition held at small scale and broke at large: two hundred contested cases a week.
- Dispute volume came in at twelve times projection. Fifty modeled. Six hundred actual.
- Three reps quietly negotiated unauthorized extensions.
Every one of those accounts counted toward percent migrated. All of them had signed. The accounts with the side deals had signed twice, once officially and once in a conversation nobody logged.
Percent migrated was not wrong. It was accurate and useless, which is worse than wrong, because nobody argues with it.
Split it four ways
A blended exception number is a smoke alarm without a location. The four types each point at a different broken thing:
| Exception type | What it means |
|---|---|
| Contested cases | The unit definition is wrong |
| Off-plan discounts | The price, or the segment value story, is wrong |
| Manual credits | The measurement is wrong |
| Side agreements | Sales does not believe the model will hold |
Our two hundred contested cases a week were a definition problem, not a price problem. Discounting harder would have made it worse.
Expect it to rise first
A rising exception rate in the early weeks is the model meeting reality while changes are still cheap. Suppress it and the problem does not go away, it relocates to renewal and arrives as churn instead of as a conversation.
What to watch for is a rate that stays flat and high after two definition fixes. That is the point where exceptions stop being about the model and start being about the comp plan, and no amount of enablement moves an incentive.
The one thing to do this week: pull your last pricing change and count the exceptions granted in the ninety days after it. If nobody was counting, that is the finding.