epic-framing

Write an epic with three parts: a Why naming the shared outcome in outcome terms, a rollup acceptance line describing the integrated behavior the child features sum to, and sequenced children each with a one-line ordering rationale based on dependency, value, or risk.

Apply the coherence test: the child features must share one outcome such that delivering some but not all still moves that outcome. If they share only a topic like billing with no common outcome, split them into separate features, bugs, or their own coherent epic.

An epic is a mid-tier grouping: coarser than a feature because it needs decomposition to deliver, and finer than an initiative because it is a specific slice of a strategic bet rather than the whole bet. Initiative-sized work should be framed with roadmap framing instead.

Split an epic when it would take more than a couple of cycles, has 15 or more children, or when no single Why covers all children. Split along the sub-outcome seam rather than arbitrarily, and never force an epic over independent features sharing only a tag.

That is a status check, not an acceptance line, because children can all be done while the integrated behavior still fails. A rollup acceptance line names the aggregate end-to-end condition that proves the bet landed, revealing missing children when it stays false.