Routine cases are the easy part of automation. The real test of a process is how it handles the exceptions, the cases that do not fit the rules.
Expect that exceptions will happen
Every process has unusual cases. Plan for them at the design stage rather than discovering them in production, so that they have a defined route instead of an improvised one.
List the likely exceptions and decide who handles each.
Make exceptions visible
Route unusual items to a clear queue with the context needed to decide quickly. Hidden or silent failures are far more costly than visible ones.
Good visibility keeps small issues small.
Give people the right information
When a case is handed over, include what the system did, what it could not determine and what is needed next. People should not have to start from scratch.
This keeps resolution times short and customers informed.
Use exceptions to improve the process
Review exception patterns on a regular basis. Frequent ones may justify a new rule, a better form or a change upstream.
Over time, the process becomes more robust and the exceptions become rarer.
- Design the exception path from the beginning.
- Surface unusual cases clearly.
- Hand over with full context.
- Learn from exception patterns.
