When the link breaks, the rule doesn’t follow
- Kind
- Day job
- Early adopters
- 15, all kept
The system
In the platform, rules attach to a record’s category, and related records get their category by being linked to each other automatically.
One source usually carries the true category; others carry none, or a placeholder. When the automatic linking fails, the rule reaches only some of the records users expect it to cover, and they find out when a record doesn’t behave the way they were promised.
Link breaks: the rule reaches 2 of 4 records.
Why it mattered
It sat in a core workflow, so one failure pattern could reach many customers at once, and the customers who had trusted the new platform first would feel it first.
The response, in three tiers
- Tier 1 By hand Fix every reported case by hand, immediately.
- Tier 2 Scripted Run a scripted fix for key accounts and the most demanding customers.
- Tier 3 Root cause Ship the root-cause fix a week later, in a maintenance release.
The thinking
Contain it in proportion to the blast radius, protect the customers who matter most first, and fix the root cause fast.
I didn’t try to fix everyone at once. Key accounts and the most demanding customers got the scripted fix first; everyone else was covered case by case until the root-cause fix shipped a week later, in a planned maintenance release rather than an emergency patch with risks of its own.
Outcome
All 15 retained, and less recurring Support and Development work.
See alsoHelping retain a $10M+ account, another day-job capability.