Workflow design
9 minutes
Designing a hand-off that doesn’t drop work
Every dropped request we have investigated failed at a hand-off. Here is the anatomy of one that holds.
Teo Lindqvist
Product Lead, Workflows
When we trace a late or lost request back through its history, the failure almost never happens inside a team. It happens between teams — in the moment one group believes it has passed the work on and the next group has not yet picked it up.
A hand-off is a contract, not a message
A message says ‘over to you’. A contract says what is being handed over, who now owns it, what ‘done’ looks like and when it is due. The difference sounds bureaucratic until you count the requests lost in the gap.
The four fields every hand-off needs
A single named owner — not a team, a person or a queue with a rota.
The minimum context to act without asking a question.
An explicit due time that starts the moment ownership changes.
A visible acknowledgement, so the sender knows it landed.
Make the acknowledgement automatic
The cheapest reliability improvement we know is an acknowledgement step with a timer. If nobody accepts the hand-off within the window, it escalates. In Kilter this is a single Review station with an SLA, but the idea works on paper too.
Design for the exception, not the happy path. The requests that matter most are the ones that arrive late on a Friday with half the information missing.
The Operator’s Ledger
One considered note every other Thursday.
Practical writing on running calmer operations. Unsubscribe in one click.
← Previous
The hidden cost of the status meeting
Next →
