Scaled Agile

Portfolio Epic Stop Criteria: Evidence Beyond Sunk Cost

Define epic stop criteria before investment and use MVP, economic, strategic, risk, and delivery evidence to continue, pivot, pause, or stop.

Portfolio Epic Stop Criteria: Evidence Beyond Sunk Cost

Stopping is a portfolio capability

Portfolio Kanban and Lean Startup practices allow epics to be removed when evidence no longer supports investment. Yet organizations often treat approval as a permanent commitment. Past spend, executive sponsorship, public roadmaps, and team identity make stopping emotionally and politically difficult. Explicit criteria make future choices less dependent on who defends the work.

Define criteria before the evidence arrives

Evidence classContinue conditionStop or pivot signal
Customer benefitLeading behavior reaches a worthwhile rangeNeed or adoption remains weak
EconomicsFuture benefit exceeds remaining cost and alternativesOpportunity cost dominates
StrategyEpic advances current themes and visionStrategy or market context changed
FeasibilityNFR and delivery risks are boundedTechnical or regulatory condition invalidates approach
FlowMVP learning arrives within agreed timeAgeing and dependencies consume capacity without learning

Separate four portfolio decisions

Continue means evidence supports the current direction. Pivot preserves the outcome while changing customer, solution, channel, architecture, or business model. Pause retains an option while waiting for named evidence or timing. Stop ends investment and initiates safe closure. Calling an indefinite pause a decision hides WIP and keeps sunk cost alive.

Run the evidence review without advocacy theatre

  1. Restate the original hypothesis and thresholds.
  2. Review supporting, contrary and segment-specific evidence.
  3. Exclude costs already spent from the forward economic comparison.
  4. Compare the epic with current alternatives and capacity constraints.
  5. Record the choice, dissent, transition actions and reconsideration trigger.

Close an epic responsibly

  • Protect customers, operations, data and regulatory obligations.
  • Capture reusable learning and architecture assets.
  • Remove or archive downstream backlog and roadmap commitments.
  • Release people and funding through a stable transition.
  • Communicate why evidence changed the decision without blaming the team.

Leading SAFe certification training develops Lean-economic and portfolio decision skills. SAFe POPM training helps product roles connect MVP and benefit evidence to backlog and roadmap changes.

A portfolio that never stops an epic is not necessarily selecting perfectly. More often, it is treating approval as success. Track stopped investment, time to decisive evidence, and resources redirected to stronger options as signs that the portfolio can learn.

Worked stop decision: assisted claims

An epic predicts that automated claims assistance will reduce handling time without increasing error or customer complaints. The MVP improves speed for simple claims but increases exceptions and specialist review for complex cases. A forward-looking comparison shows that broad rollout costs more than its remaining benefit. Leaders stop universal implementation, preserve the simple-claim capability, retire unused integration work, and fund a narrower discovery option for complex claims.

Design psychological safety around stopping

Teams will hide weak evidence if stopping is interpreted as failure or career damage. Leaders should reward early learning, accurate risk reporting, and responsible closure while remaining accountable for repeated weak decisions. Separate the quality of the original decision under available evidence from the eventual outcome. This encourages honest thresholds without removing scrutiny from investment choices.

Questions that expose sunk-cost reasoning

  • Would we start this epic today with current evidence?
  • Which future benefit depends on money already spent?
  • What better option is displaced by continuing?
  • Are we preserving a customer outcome or defending a chosen solution?
  • What transition cost is real, and what cost is only emotional or political?