A quarterly plan allocates 100% of engineering capacity to roadmap features. The rationale is that maintenance work can be picked up as it arises. Review the plan.

What is actually required

The plan is not 100% roadmap; it is roadmap plus an unmeasured amount of unplanned work, with the second part hidden. Dependency upgrades, certificate and credential rotation, cloud deprecation notices, security patches, incident follow-ups and support escalations arrive whether or not they were planned. A plan without a line for them does not remove the work; it removes the visibility of the work, so the roadmap simply delivers late and nobody can say why.

What I would change, and why it is safe to

Name the baseline and reserve it. Organisations that measure this typically find unplanned and keeping-the-lights-on work consuming somewhere between 20% and 40% of capacity, and a team with no reservation will spend that anyway, from the roadmap's allocation, invisibly.

Reserving it is safe because the total output does not change: the work happens either way. What changes is that the roadmap commitment becomes achievable, and the trade-off becomes visible to the people making commitments.

The one change that matters

Publish what the reservation was spent on, each quarter. Without this, the reservation is perceived as slack and is reclaimed at the first pressure. With it, the conversation becomes evidence-based: a quarter spent mostly on a forced cloud migration is a fact leadership can act on, perhaps by funding the automation that reduces it.

What I would leave alone

Do not try to eliminate the category. A proposal to drive unplanned work to zero produces either very slow change or a freeze on dependency updates, and the second is how a system arrives at an unsupported language runtime three years later, which costs multiples of the upgrades it avoided.

How to argue this in the review

Not as "we need slack", which sounds like asking for less accountability. As forecast accuracy: the last four quarters delivered a measurable fraction of what was committed, the gap was consumed by work that was never in the plan, and the proposal is to put it in the plan so the commitment means something. The number from the team's own history is the argument; the principle alone will not survive a budget conversation.

When not to reserve capacity

In a genuine, time-boxed push — a funding deadline, a regulatory date, a peak event — for one quarter, with the deferred maintenance written down and scheduled. Declared, bounded and recorded, it is a legitimate decision. Repeated for a second and third quarter, it becomes the mechanism by which systems become unsupportable, and the honest thing to say at the second request is that the bill is now accruing interest.