Triggers
Any one can start it
- Usage −50% over 12mo with no recovery
- Maintenance cost > value delivered
- Conflicts with strategy
- Unsafe tech debt
- A better alternative exists
Execution, Lifecycle & Sunsetting
Shipping is the start of a product's life, not the finish — the hardest, most valuable PM call is deciding when to kill it.
Discovery runs parallel to delivery, every product ages through the lifecycle, and a disciplined sunset is a brand decision. Read the 2×2, then act.
Growing usage · low fit
Harvest
Milk the cash flow; don't reinvest.
Growing usage · high fit
Reinvest
Double down — it's working and it matters.
Declining usage · low fit
Kill
Retire it in stages; free the capacity.
Declining usage · high fit
Migrate
Move users to a better vehicle; reinvent.
Sunset-or-double-down · usage trend × strategic fit
Above the fold
The three execution moves that matter most.
01
Run discovery parallel to delivery.
Time-box ~20–30% of PM time every week; findings feed the backlog 1–2 sprints ahead, never the current sprint. Never trade this week's customer interview for a meeting.
02
Manage the whole lifecycle.
Intro → Growth → Maturity → Decline, each with its own move. Run age-structure analysis — a healthy portfolio holds products in every stage.
03
Killing is harder than launching.
The champion has ego invested and the few remaining users are vocal. Decide with the sunset-or-double-down 2×2 and frame the call as opportunity cost.
Dual-track agile · Torres / Cagan
Discovery runs parallel to delivery.
Time-boxed ~20–30% of PM time every week — not project-based. The trio (PM + design + eng) discovers together.
Discovery track~20–30% / week
Continuous, weekly, trio-run. Findings feed the backlog 1–2 sprints ahead — never the current sprint.
Delivery trackthe current sprint
Builds what discovery has already de-risked. Never trade this week's customer interview for a meeting.
If discovery keeps losing to delivery urgency
That's a leadership problem, not time-management. AI-era parallel: the 5-day Proto-Cycle + Shipyard 6-function concurrency (the Udezues, Building Rocketships 2025) — prototype-as-requirements, ship-to-learn in days not sprints. Hedged 2025 · cadence illustrative
Product lifecycle management · Aumayr
Every product ages through four stages.
Run age-structure analysis — a healthy portfolio holds products in every stage.
Intro
Educate
Early adopters; premium price OK.
Growth
Differentiate
Capture share, expand distribution.
Maturity
Defend
Retain, segment, cost-optimize; defend via service / ecosystem.
Decline
Harvest
Harvest / consolidate / reinvent.
New-product development · Cooper Stage-Gate (pointer)
Gate NPD on a differentiated, superior product.
The #1 success driver is a differentiated, superior product with sharp early definition — gate on that. “Stage-Gate” is a trademark.
1
Idea
2
Scope
3
Business case
4
Develop
5
Test
6
Launch
Score concepts with NewProd ~73–84% pre-Development accuracy · directional; rank by Productivity Index (NPV ÷ constraining resource, go-forward costs only); fund Strategic Buckets, not one-off projects. Agile-Stage-Gate suits physical products (documented challenges, not a clean fit).
Deprecation & sunsetting
The political reality is the point.
Your sunset quality reflects your brand. Tell customers early with a migration path, retire in stages, and feed the post-mortem back into strategy.
Notice
Tell customers early
- ≥6mo enterprise / ≥3mo consumer
- Ship a migration path and tooling
- Name the sunk cost to leadership, then move on
Decide which
Sunset-or-double-down 2×2
- Usage trend × strategic fit
- → reinvest / harvest / migrate / kill
Staged retirement
Stage 1
Stop new signups
Stage 2
Freeze updates
Stage 3
Kill
Stage 4
Post-mortem
Back into strategy.
Calibration trap · killing is harder than launching
The champion has ego invested and the few remaining users are vocal — so a dying product survives on sunk cost. Frame the call as opportunity cost: “an engineer maintaining X is one not building Y.”