Insights
What ongoing care for a workflow should and shouldn't cover
Once a workflow is running, the question changes. It is no longer whether it works, but who keeps it working, and how anyone would know when it stops. Without a clear answer, responsibility for exceptions, dependencies and changes spreads thin and becomes hard to verify.
What care should cover
Ongoing care is useful when its contents are specific. It should include:
- Agreed checks: what is looked at, and what a healthy result looks like.
- Issue review: how a problem is noticed, recorded and examined, and who decides what happens next.
- A record of exceptions and maintenance, so the history of the workflow is not held in one person's memory.
- Bounded maintenance: the routine upkeep the workflow needs to keep running as agreed.
- Improvements proposed from evidence of how the workflow actually runs, not from a list of features.
What care should not quietly become
Care has edges, and they should be stated as clearly as its contents. Care for a workflow is not round-the-clock monitoring, emergency response, unlimited repair or a security service unless those are agreed separately. It is also not a substantial rebuild carried out under the name of maintenance.
When a workflow needs a larger change, that change should be decided and scoped on its own. Folding it into care hides the decision and makes both harder to judge.
Start with a review
Before agreeing what care covers, review the workflow: its tools, its permissions, its dependencies and its known failure paths. Care should cover only workflows that have been reviewed and accepted, because a workflow nobody has examined cannot be cared for responsibly.
The review may show that the workflow needs to be stabilised before care makes sense, that a separate change should come first, or that no care is needed at all. Deciding that no care is needed is a complete outcome.
Questions to ask of any care arrangement
- Exactly which workflows are covered, and which are not?
- What is checked, and what counts as a problem?
- Who sees an exception first, and where is it recorded?
- Where does care end, and where does a separately scoped change begin?
- Who decides which improvements are made?