GF · Greenfield
Tradeoff: fastest to iterate, no legacy constraints — but rebuilds what exists and carries naming/discovery debt.
- [what's new]
- + [win: no legacy coupling]
- − [cost: duplicate of existing thing]
BF·Backend-first
Tradeoff: extends the existing contract, preserves data lineage — but the change is bigger and slower to demo.
- [what lands first]
- + [win: reuses the store]
- − [cost: needs migration/flag]
Hybrid
Tradeoff: new path for the new case, existing path for the old — most work now, least throwaway later.
- [routing decision]
- + [win: no breakage]
- − [cost: two paths to learn]
Decide with: [owner / group] — before [phase gate]. Unresolved: [the open question that changes the pick].
Picked . This throwaway file is done — record the pick in the PRD / bead, then discard this artifact.