PI Planning in SAFe: Agenda, Roles, Outputs and Common Mistakes
PI Planning is the cadence-based event at the centre of the Scaled Agile Framework (SAFe®). It brings an Agile Release Train (ART), often a hundred people or more, together to agree what it will deliver in the next Planning Interval (PI) and how its teams depend on each other. The event matters because most large-scale delivery problems are alignment problems: dependencies discovered late, priorities that differ by team and plans nobody believes.
This guide is for people preparing to take part in, facilitate or sponsor PI Planning. It explains what the event is for, who takes part, the typical two-day agenda, what teams leave with, how to prepare, how to run it remotely, the common mistakes and what happens after the event. For a short definition, see our glossary entry on PI Planning.
A note on terms
In earlier versions of SAFe, "PI" stood for Program Increment. SAFe 6.0 renamed it the Planning Interval, and several related terms changed at the same time, for example the "program board" became the ART planning board. You will see both sets of terms in organisations and older materials. This guide uses the current terms.
What PI Planning is for
A Planning Interval is a timebox, typically eight to twelve weeks and usually four or five iterations, in which an ART delivers incremental value. PI Planning has four practical purposes:
- Shared understanding of the business context, vision and priorities.
- Team plans for the iterations ahead, built by the teams themselves.
- Visible dependencies and risks, found while the people involved are in the same conversation.
- Commitment from teams and business owners to a set of objectives.
It replaces the long chain of emails, meetings and approvals that large programmes often use to coordinate. Instead, the people who do the work make the plan together, with the people who make trade-offs in the room.
Why face-to-face planning works
The value comes from compressing months of coordination into two days. When a team discovers it needs an interface from another team, it walks over (or opens a breakout room) and agrees it then and there. When the plans show more work than capacity, leaders decide what to cut that evening, not three weeks later. The event is expensive in people's time, and it pays back by avoiding the much larger cost of misalignment: blocked teams, rework and features that arrive without the pieces they depend on.
Who takes part
| Role | Contribution |
|---|---|
| Agile teams (Developers, Scrum Masters/Team Coaches, Product Owners) | Draft plans, estimate capacity, identify dependencies and risks |
| Product Management | Present the vision and the prioritised features for the PI |
| System Architect | Explain architectural guidance, enablers and constraints |
| Release Train Engineer (RTE) | Facilitates the event and the ART |
| Business Owners | Explain business context, review team plans and assign business value to objectives |
| Shared services and other stakeholders | Clarify needs, constraints and decisions |
Everyone on the train attends. That is deliberate: plans made by representatives miss what the people doing the work know. It also builds commitment. People are far more likely to deliver a plan they helped create than one handed to them, and they understand the reasons behind the trade-offs because they saw them made.
A typical agenda
SAFe publishes a standard two-day agenda. The detail can change between releases, so treat the shape below as typical and check the current guidance.
Day 1
- Business context. A senior leader describes the current state of the business and how the ART's work supports it.
- Product or solution vision. Product Management presents the vision and the top features for the PI.
- Architecture vision and development practices. The System Architect describes architectural guidance; a senior leader may cover practices such as quality and DevOps.
- Planning context. The RTE explains the planning process and expected outputs.
- Team breakouts (1). Teams estimate their capacity for each iteration, identify the backlog items they can deliver, draft PI objectives and list risks and dependencies.
- Draft plan review. Each team presents its draft plan, objectives, risks and dependencies to the whole train.
- Management review and problem-solving. After the draft review, leaders meet to resolve the problems the plans revealed: scope, people, priorities and constraints.
Day 2
- Planning adjustments. Leaders present the decisions from the management review.
- Team breakouts (2). Teams adjust their plans, finalise objectives, and Business Owners visit teams to assign business value to each objective.
- Final plan review. Each team presents its final plan, and the train confirms the ART planning board.
- ART risks. Risks are discussed openly and categorised using ROAM.
- Confidence vote. Everyone votes, typically on a scale of one to five fingers, on how confident they are that the plan is achievable.
- Plan rework if necessary. If confidence is low, teams and leaders rework the plan until the vote shows reasonable confidence.
- Planning retrospective and moving forward. The train reflects on the event itself and confirms next steps.
What teams leave with
- Committed PI objectives for each team, written in business language, with business value assigned by Business Owners. Some objectives are marked uncommitted: the team will try to deliver them but does not commit, which gives a buffer for uncertainty.
- An ART planning board showing features, milestones and dependencies between teams across the iterations.
- A risk list categorised with ROAM: Resolved, Owned, Accepted or Mitigated.
- A confidence vote result showing how comfortable the group is with the plan.
- Iteration plans for each team, at least for the first iteration.
These are not decoration. The ART planning board and objectives are what teams manage against during the PI, and the risks are what the RTE keeps in view.
Writing good PI objectives
| Weak objective | Better objective |
|---|---|
| "Build feature 42" | "Customers can reset their own password without calling support" |
| "Finish API work" | "Payments team can test against the new account API by iteration 3" |
| "Improve performance" | "Checkout page loads in under two seconds for 95% of users" |
Good objectives describe outcomes that Business Owners can recognise and value. They help everyone understand why the work matters.
How to prepare
PI Planning succeeds or fails before it starts. Preparation usually begins weeks in advance.
Content readiness
- Refine and prioritise the ART backlog so the top features are understood, sized and have clear acceptance criteria.
- Share the vision and architecture guidance before the event, so teams arrive informed.
- Check team capacity: holidays, support duties, training and other commitments.
- Identify known dependencies with other trains, vendors and shared services, and invite the right people.
People readiness
- Brief Business Owners on their role: explaining context, assigning business value and taking part in the management review.
- Brief leaders that they must make trade-off decisions during the event, not after it.
- Prepare Scrum Masters/Team Coaches to facilitate team breakouts.
Logistics
- Book space, or set up virtual rooms, for every team plus a main room.
- Prepare the ART planning board, physical or digital.
- Plan the timetable, including breaks and buffer time.
- Test tools in advance if any part is remote.
Running PI Planning remotely
Many ARTs now run PI Planning fully or partly online. It can work well, with adjustments:
- Split across more days. Shorter days, spread over three or four days, reduce screen fatigue, especially across time zones.
- Use breakout rooms for each team and a clear way to visit other teams' rooms.
- Use a digital ART planning board that everyone can see and edit.
- Assign clear roles: who facilitates, who keeps time, who manages the board and who captures risks.
- Plan time zones carefully. If teams span many zones, rotate the inconvenience fairly.
- Keep cameras and conversation active. Remote events that turn into webinars lose the value of face-to-face planning.
Common mistakes
- Arriving unprepared. If the backlog is not refined, the event becomes backlog refinement and runs out of time.
- Treating objectives as a to-do list. Objectives should describe outcomes in business language, not repeat feature names.
- Planning to 100% capacity. Teams that ignore holidays, support load and uncertainty will miss.
- Hiding dependencies. If a dependency is not visible on the board, it will surface as a surprise.
- Management absence. Leaders must be present to make trade-offs. If they are not, problems are escalated rather than solved.
- Ignoring the confidence vote. A low vote is information, not an inconvenience.
- Skipping the retrospective. Each PI Planning should be better than the last.
- Turning remote events into a webinar. Remote events need small-group tools, clear roles and breaks.
Tips for first-time participants
- Know your team's capacity before you arrive, including holidays and support duties.
- Read the top features in advance and note questions for Product Management.
- Speak up about dependencies early. The draft plan review is the cheapest place to find them.
- Write objectives the business understands. If a Business Owner cannot see the value, rewrite it.
- Vote honestly. A confidence vote of two with a reason is more useful than a polite four.
After PI Planning
The plan is a starting point, not a contract. During the PI:
- Teams run their iterations and update their plans as they learn.
- The RTE and Scrum Masters/Team Coaches hold regular ART sync meetings to track progress, dependencies and risks.
- The train shows integrated work at System Demos at the end of each iteration.
- The Innovation and Planning (IP) iteration at the end of the PI gives time for innovation, learning, infrastructure work and preparation for the next PI Planning.
- The PI ends with Inspect and Adapt, where the train reviews results, measures performance against objectives, and solves its biggest problems.
A short example
A bank's digital channels ART has nine teams. At its first PI Planning, the draft plan review on day 1 shows that four teams depend on a new customer-profile service that one team cannot build until iteration 4. The management review that evening agrees to move two developers from a lower-priority feature to the profile team and to deliver a simplified version of the service by iteration 2. On day 2, the dependent teams re-plan around the earlier date, and the risk of the full service slipping is marked as owned by the profile team's Product Owner.
The confidence vote averages just above three. Without PI Planning, the train would probably have discovered the dependency in iteration 3, when four teams were blocked at once.
Measuring whether PI Planning is working
| Measure | What it tells you |
|---|---|
| Confidence vote trend | Whether teams believe their plans |
| Predictability (business value achieved against planned) | Whether the train delivers what it commits |
| Late-discovered dependencies | Whether planning is surfacing dependencies early enough |
| Unplanned work during the PI | Whether capacity planning is realistic |
| Retrospective actions completed | Whether the event itself is improving |
Where to learn more
PI Planning is covered in SAFe's Leading SAFe course, and in more depth for those who facilitate it, such as Release Train Engineers and Scrum Masters. See CertScope's Leading SAFe® training and SAFe® Practice Consultant (SPC) training, and compare routes in our SAFe certification path guide. If your question is about how programmes differ from projects and portfolios, read project, programme and portfolio management explained.
Get free certification pathway advice →
Key takeaways
- PI Planning aligns an Agile Release Train on objectives for the next Planning Interval.
- SAFe 6.0 renamed Program Increment to Planning Interval.
- The event produces committed objectives, an ART planning board, ROAM-categorised risks and a confidence vote.
- Most failures start with poor preparation or absent leaders.
- Remote PI Planning works with shorter days, breakout rooms and clear roles.
Frequently Asked Questions
How long is PI Planning?
SAFe describes PI Planning as a two-day event. Remote and hybrid trains often spread it over three or four shorter days.
Who must attend?
Everyone on the Agile Release Train: the teams, Product Management, the System Architect, the Release Train Engineer, Business Owners and other key stakeholders.
What is a PI objective?
A short statement, in business language, of what a team intends to deliver in the Planning Interval. Business Owners assign each objective a business value.
Is PI Planning the same as sprint planning?
No. Sprint planning plans one team's iteration. PI Planning plans a multi-iteration Planning Interval across many teams and their dependencies.
Next step
If you are preparing for your first PI Planning, tell us your role and your organisation's SAFe experience. We can suggest which course will help most.
Ready to get certified?
Browse our full catalog of expert-led courses and certifications.