SAFe LPM Refresher
The Epic Hypothesis Statement
- For (target customer/user)
- Who (problem/opportunity)
- The (proposed solution)
- Is a (solution type)
- That (expected benefit)
- Unlike (competitor/current solution)
- Our solution (the differentiator/why this approach is better)
Making Hypotheses Testable
- Minimum Viable Product (MVP): An Epic in SAFe has a specifically defined MVP, which is not a full product but the smallest functional version of the solution needed to test the hypothesis with real users. This limits risk and allows for early learning.
- Business Outcomes: These are the measurable benefits the business expects to achieve if the hypothesis is correct (e.g., “reduce support calls by 30%”).
- Leading Indicators: These are early metrics that help predict whether the business outcome is likely to be achieved before the full solution is deployed. Examples include the number of users visiting a new feature’s information page or time spent on a specific task.
Developing an MVP
A core component of Lean Startup methodology is the build-measure-learn feedback loop. The first step is figuring out the problem that needs to be solved and then developing a minimum viable product (MVP) to begin the process of learning as quickly as possible. Once the MVP is established, a startup can work on refining the solution. This will involve measurement and learning and must include actionable metrics that can demonstrate cause and effect.
An MVP-scoped epic is the smallest explicitly funded investment slice that is expected to generate decisive learning about a business hypothesis within a bounded time horizon.
Epic Workflow and Decision Points
- Funnel/Review: The idea is initially captured and discussed.
- Analyzing: The Epic Hypothesis Statement is drafted, the Lean Business Case is developed, and the MVP is defined.
- Implementing: Features related to the MVP are developed by Agile Release Trains (ARTs).
- Evaluating: The MVP is released, and leading indicators are tracked to gather empirical data.
- Pivot or Persevere: Portfolio leadership uses the data and leading indicators to decide whether to continue investing (persevere), change direction (pivot), or stop the initiative altogether.
What You Are Actually Managing in LPM
You are managing a bounded investment decision. That investment decision is best represented as:
“The smallest funded slice that will allow us to validate or invalidate a meaningful hypothesis.”
That is your MVP.
How MVP Fits with Epic Aging – What To Track
You track aging from the moment the MVP slice is pulled into implementation (ART Backlog discovery & delivery), not from the moment the idea was conceived.
In other words: Epic aging after pull = investment exposure.
At the portfolio level, we don’t track epics as indefinite initiatives. We track MVP-sized investment slices with explicit learning goals. That’s what makes WIP limits and aging meaningful.
If an epic cannot finish, it cannot age—and if it cannot age, it cannot be governed by flow.
So, how long do we wait to make a pivot vs persevere decision? That finish line is a decision point, not “delivery completed”.
Why 2 PIs Works So Well
Why ~2 PIs is a very strong default (with nuance). An epic sized to produce learning within ~2 PIs:
-
Preserves economic optionality
-
Keeps cost of delay visible
-
Allows aging to signal risk early
-
Prevents portfolio WIP from calcifying
-
Makes stopping or pivoting politically survivable
Crucially: Beyond ~2 PIs, aging stops being a leading indicator and becomes a post-mortem.
At that point, portfolio governance drifts back to calendar reviews and sunk-cost thinking.
Why 1 PI Is Often Too Small
At the portfolio level:
-
Some strategic bets need cross-ART coordination
-
Some learning requires real market exposure
-
Some architectural moves need runway
A 1-PI limit can:
-
Bias toward incrementalism
-
Kill Horizon 2 / Horizon 3 work
-
Encourage trivial MVPs that don’t reduce real uncertainty
That’s why ~2 PIs is the sweet spot:
-
Long enough for meaningful learning
-
Short enough to preserve flow control
- Epics should be scoped so that decisive learning is expected within roughly two PIs.
Should Portfolio Review Look Like ART Sync?
Yes—structurally the same, but economically richer.
Portfolio review should absolutely include:
-
A list of in-progress epics
-
Ranked by age since pulled into implementation
-
With explicit aging thresholds highlighted
What Portfolio Aging Actually Measures
At the portfolio level:
-
Epic age starts when the epic is approved and pulled into implementation
-
Epic age ends when the MVP learning objective is met and a decision is made
Epic aging is therefore a measure of: Investment exposure without validated learning
(That is the portfolio analogue of feature aging at the ART level).
Recommended Portfolio Aging Thresholds
We don’t want many thresholds—just enough to drive attention.
A simple model:
-
Expected learning window: ~2 PIs
-
Early warning threshold: ~2.5 PIs
-
Escalation threshold: ~3 PIs
Crossing thresholds does not mean failure.
It means governance attention is required.
Portfolio Review Agenda (Flow-Based)
A healthy portfolio review agenda looks very familiar now:
-
Portfolio WIP vs limit
-
In-progress epics ranked by age
-
Epics beyond aging threshold
-
Evidence gathered so far
-
Explicit decisions:
-
- Continue
- Pivot
- Stop
- Re-slice
And—this is key— If no epic has crossed a threshold, the meeting should be short.
Silence means the system is working.
LPM Portfolio Review vs ART Sync (vs Team Sync)
The structure is the same.
The decision content is different.
| Level | Aging Signals | Decisions |
|---|---|---|
| Daily Standup | Story aging | Swarm to finish, unblock dependencies, reduce scope/re-slice |
| ART Sync | Feature aging | Unblock, de-scope, resequence |
| Portfolio Review | Epic aging | Continue, pivot, stop, re-fund |
Same mechanics.
Different economic altitude.
This consistency is powerful—it teaches the organization one control model, reused everywhere.
-
Stories age in days
-
Features age in sprints
-
Epics age in PIs
Same signals.
Same governance logic.
Different time horizons.
