Lean Portfolio Management in Practice

A step-by-step guide to connecting strategy, budget, and delivery without turning the portfolio into another layer of bureaucracy.

It’s December. Leadership locks next year’s budget in a room, with a spreadsheet of 80 approved projects. Each one has a scope, a deadline, and a cost, all based on what people knew in October. By March, half of that spreadsheet no longer makes sense, but changing anything means another approval cycle. Teams are spread across four or five initiatives at once. Everything is “in progress,” almost nothing gets finished, and the monthly portfolio meeting has turned into a reading of green, yellow, and red status reports.

If this sounds familiar, the problem probably isn’t the teams. It’s how the portfolio decides what starts, how much funding it gets, and when it stops.

Lean Portfolio Management, or LPM, is the most structured answer to that problem. SAFe organized it, but it also works in companies that don’t use the framework. In this article I’ll explain what it is, who takes part, and, above all, how to make it work: an eight-step roadmap, a 90-day plan, and the mistakes I see repeated most often. The goal is for a leadership group to follow this text and come out with LPM running, not just understood.

What LPM is, in one sentence

LPM is a way to decide, fund, and track a set of initiatives by looking at the flow of value to the customer instead of at isolated projects.

In SAFe, it is one of the seven core competencies of business agility and rests on three dimensions:

• Strategy and investment funding: deciding where the portfolio is going and how much money each value stream gets.
• Agile portfolio operations: keeping work flowing across value streams, with coordination and shared practices.
• Lean governance: controlling spending, measuring results, forecasting costs, and handling compliance and audit without blocking delivery.

One shift in mindset runs through all three: moving from “approving projects with fixed scope” to “funding value streams and betting on hypotheses that can be tested.”

Who takes part

LPM is a function, not a job title or a new committee. In practice, it brings together people who already exist in the company but usually decide separately:

• Business executives and Business Owners: accountable for results; they decide strategy and investment.
• Finance: without them, LPM becomes a theoretical exercise, because the budget keeps being built the old way.
• Enterprise Architect: keeps technical strategy aligned with business strategy and usually owns enabler epics.
• Epic Owners: the people who carry each major portfolio initiative from idea to a validated hypothesis.
• Train and value stream leadership: RTEs, Product Management, and whoever is accountable for each value stream.
• APMO (Agile PMO), Agile Coaches, and SPCs: they facilitate the events, keep the cadence, and help the group learn a new way of deciding.
• Risk, compliance, security, and audit: in regulated industries, they join from the start, not as a gate at the end.

Before you start: three minimum conditions

LPM doesn’t need a full transformation to get started, but it does need three things. Without them, the group will spend energy on meetings that decide nothing.

1. A sponsor with power over the budget. Someone who can actually change how money is allocated. If the group’s only lever is to “recommend,” LPM becomes a PMO with a new name.
2. Value streams identified, even roughly. Knowing how value reaches the customer: account opening, lending, payments, customer service. A value stream map built with the people who work in the flow helps a lot here. It doesn’t need to be perfect; it will be refined along the way.
3. Stable teams, or trains being formed. Funding value streams only makes sense if there are long-lived teams working in them. If every project assembles and disbands a team, start there.

If your company doesn’t have all three yet, you can begin with steps 1 to 4 of the roadmap below, which work even in a project-based model, and leave budgets and guardrails for later.

What if your company doesn’t use SAFe?

You can do all of this without SAFe. I use SAFe vocabulary because it’s the most widespread in large companies and has detailed official material, but the ideas behind LPM didn’t start there. They come from Lean, from flow economics applied to product development, and from movements like Beyond Budgeting, which has challenged the fixed annual budget since the late 1990s.

Without trains or PI Planning, the eight steps still apply. The names change, not the logic.

A few approaches that fit well with the roadmap:

• Flight Levels, by Klaus Leopold: the model’s top level is precisely strategy and portfolio, focused on limiting work in progress and coordinating what happens between teams.
• Beyond Budgeting: for anyone who wants to change the budget cycle and replace fixed annual targets with more frequent reviews.
• OKRs with quarterly reviews: they play the role of strategic themes and the strategic review without requiring anything else.
• Cost of delay: prioritizing by dividing the cost of waiting by the duration works in any model. WSJF is SAFe’s version of this idea.

With or without SAFe, there are four things you can’t skip: a group with real power over the budget, an honest list of what’s in progress, a limit on starting new things, and a cadence that ends in decisions.

The eight-step roadmap

The steps are in order, but they’re not a waterfall. The group will come back to each of them many times. What matters is finishing the first full cycle, even if it’s imperfect, and improving from there.

Step 1. Form the group and agree on how it decides

Goal: a small group with real decision power that knows what it decides on its own and what it escalates.

How to do it:

• Start small: five to nine people usually works. A group that’s too large doesn’t decide; it presents.
• Write a one-page LPM charter: purpose, members, what the group decides, cadence, and how a decision is made when there’s no consensus.
• Make explicit what leaves the group’s hands: decisions below a certain amount stay with the value streams. That’s decentralization with clear limits, and it’s the foundation for the guardrails in step 6.

Sign that it worked: someone outside the group can answer “who decides whether this initiative starts?” without having to ask.

Step 2. Take a picture of the portfolio as it is today

Goal: a single, one-page view of what the portfolio delivers, to whom, and at what cost.

How to do it:

• Use the Portfolio Canvas, SAFe’s adaptation of the Business Model Canvas. It describes the value streams, solutions, customers, channels, partners, key activities and resources, cost structure, and revenue or benefits.
• Fill it in during a half-day workshop with the LPM group and representatives of each value stream. Use real cost and revenue data, even if approximate.
• After the current state, sketch the future state: where the portfolio should be in 12 to 24 months. A SWOT analysis helps get there.

Sign that it worked: the group discovers at least one thing it didn’t know. For example, a value stream that consumes a lot of budget and delivers little, or a customer nobody was considering.

Step 3. Turn strategy into themes and OKRs

Goal: three to five strategic themes that link company strategy to the portfolio, each with measurable results.

How to do it:

• Strategic themes are the business objectives that set the portfolio apart in the period. They come from comparing the current canvas with the future one.
• Write each theme as an OKR: one qualitative objective and two to four key results with a number, a baseline, and a deadline.
• Keep the number small. If everything is a strategic theme, nothing is.

Sign that it worked: any new epic can be compared against the themes, and the group can say “that’s not in our strategy” without embarrassment.

Step 4. Stop starting, start finishing

Goal: reduce the number of initiatives running in parallel so that what matters gets done faster.

How to do it:

• List everything in progress in the portfolio: projects, programs, the “small” initiatives nobody tracks. For each one, note the owner, estimated monthly cost, linked strategic theme, and how many teams are involved.
• Classify each item into one of four decisions: continue, pause, stop, or reassess as an epic (when it makes sense, but nobody can say what result is expected).
• Set a work-in-progress limit for the portfolio. There’s no magic number; start with what you have today minus 20% to 30%, and adjust every quarter.

This step is usually the hardest politically and the one that pays off fastest. Every initiative has a sponsor, and stopping something feels like admitting a mistake. It helps to frame the decision as economics (“freeing capacity for what matters most right now”) rather than as a judgment of whoever proposed it.

Sign that it worked: teams work on fewer initiatives at once, and the time to finish the remaining ones starts to drop.

So you don’t start from scratch: I’ve put together a blank Excel portfolio cost spreadsheet with ready-made formulas. It calculates the cost of each team and each initiative, flags what goes over the epic threshold or has no strategic theme, compares planned and actual budget, and shows whether investment allocation respects the guardrails. It works with or without SAFe.

Download the spreadsheet here.

Step 5. Set up the Portfolio Kanban

Goal: make every major initiative follow the same visible path, from idea to the decision to continue or stop.

How to do it:

• Define what a portfolio epic is: an initiative big enough to cross an agreed threshold (of cost, duration, or impact on several value streams). Below that threshold, the decision stays with the value stream.
• Build the board with the states SAFe suggests: Funnel, Reviewing, Analyzing, Ready, Implementing, and Done. Reviewing, Analyzing, and Ready have explicit WIP limits. The Funnel doesn’t: anyone can put an idea there.
• In Reviewing, the Epic Owner writes the epic hypothesis statement. In Analyzing, they build the Lean Business Case and define the MVP, the smallest experiment that can prove or disprove the hypothesis.
• The go / no-go decision happens at the Portfolio Sync, based on the Lean Business Case. “No” sends the epic straight to Done, and that’s a legitimate outcome, not a failure.
• In Implementing, the MVP comes first. If the hypothesis holds, the epic perseveres and gets more investment. If not, the group decides whether to pivot or stop.
• An epic is done when the hypothesis has been evaluated and it’s no longer a portfolio concern, not when the entire imagined scope has been delivered.

Sign that it worked: the group can look at the board and say, in minutes, what’s stuck, where, and why.

The epic hypothesis statement forces whoever proposes an initiative to say, before asking for money, who it’s for, what result they expect, and how they’ll know early if it’s on track.

The Lean Business Case isn’t the traditional 40-page business case. It fits on one or two pages and exists to support a decision, not to lock in a scope.

How to prioritize what gets in: SAFe uses WSJF (Weighted Shortest Job First), which puts first whatever delivers the most value per unit of effort, considering the cost of waiting. Cost of delay adds up three factors: user and business value, time criticality, and risk reduction or opportunity enablement. Scoring is relative, on a Fibonacci-like scale (1, 2, 3, 5, 8, 13, 20), comparing epics against each other.

One caveat: WSJF is a structured conversation, not a formula that decides on its own. Its value is in getting people to discuss why one item is more urgent than another. If the number contradicts the group’s judgment, discuss the scoring instead of obeying the spreadsheet.

Step 6. Fund value streams, with guardrails

Goal: replace project-by-project approval with budgets per value stream, with clear rules for how the money can be used.

Get Bruno Novaes’s stories in your inbox

Join Medium for free to get updates from this writer.

Remember me for faster sign in

This is the heart of LPM and the step that depends most on Finance. In the traditional model, each project competes for money, assembles a temporary team, and is held accountable for scope and deadline. In the Lean model, each value stream gets a budget to sustain stable teams and decides, within limits, how to use it.

How to do it:

• Start with a budget per value stream that reflects the real cost of its people and operations today. Don’t try to redistribute everything in the first cycle.
• Define SAFe’s four budget guardrails: guiding investments by horizon, capacity allocation, approving significant initiatives, and continuous Business Owner engagement. The first two are quantitative; the last two define who decides and how.
• Agree with Finance on how spending will be tracked (by value stream, monthly) and on the accounting treatment. In companies that capitalize software development, this needs to be settled before, not after.

Sign that it worked: a value stream can change an initiative’s priority within its budget without asking for approval, and everyone knows exactly when they do need to ask.

Participatory Budgeting. Redistributing the budget across value streams happens in a dedicated event, Participatory Budgeting. Instead of each leader defending their slice in a room, people work in small groups, each with part of the total budget, and decide together where to invest. The groups’ results are consolidated, and LPM makes the final decision. The most valuable effect is that every leader starts seeing the whole portfolio, not just their own area.

Step 7. Get the cadence running

Goal: a few events, each with a clear purpose, that replace status meetings.

SAFe proposes three LPM events:

• Portfolio Sync, usually monthly: tracks epic progress, resolves dependencies and impediments across value streams, and makes go / no-go decisions. The question is “is what we decided being delivered, and what’s blocking it?”
• Strategic Portfolio Review, usually quarterly: revisits the portfolio vision, strategic themes, and investments. The question is “are we investing in the right things?” In the months when it happens, it usually replaces the Portfolio Sync.
• Participatory Budgeting, usually once or twice a year: redistributes the budget across value streams.

Align the calendar with the trains’ planning (PI Planning): the strategic review should come before planning, so decisions reach the teams in time.

Sign that it worked: events end with recorded decisions, not slides read aloud. The old status meeting can be cancelled and nobody misses it.

Step 8. Measure and govern without blocking flow

Goal: know whether the portfolio is producing results, whether work is flowing, and whether money is well spent, with as little reporting as possible.

How to do it:

• Organize metrics in three layers, as SAFe proposes: outcomes (the key results of strategic themes and of each epic), flow (how long an epic takes from approval to done, how many are in progress, how investment is distributed across themes), and competency (how the group assesses its own LPM maturity).
• Track the budget by value stream, comparing planned with actual, every month. The forecast is redone every cycle, not once a year.
• Bring compliance, risk, and audit into the flow: compliance criteria in the epic hypothesis and in the definition of done, evidence produced during the work instead of rebuilt at the end.
• Data comes from the tools teams already use. If building the dashboard takes a week of spreadsheets, nobody will keep it up.

Sign that it worked: the dashboard fits on one screen and answers three questions: are we getting the results we promised, is work flowing, and is the money going where we decided?

The 90-day plan

Eight steps sound like a lot, but they fit into a quarter if the group works in short cycles. The plan below organizes the steps into three phases of roughly a month each, every one ending with a concrete deliverable.

• Days 1 to 30, see: signed LPM charter, current-state Portfolio Canvas, complete inventory of everything in progress. Milestone: the first Portfolio Sync, already with decisions to pause or stop.
• Days 31 to 60, decide: strategic themes with OKRs, Portfolio Kanban up and running with the epics that survived the inventory, epic threshold defined, guardrails written with Finance. Milestone: the first Strategic Portfolio Review.
• Days 61 to 90, run: budget per value stream for the next cycle, the first Participatory Budgeting (even a scaled-down version), a three-layer dashboard, and a retrospective of LPM itself. Milestone: the next PI Planning (or quarterly planning, without SAFe) already receives priorities from the portfolio.

On day 90, don’t expect a mature LPM. Expect a first full cycle, with real decisions made, and a clear list of what to improve in the second.

The most common mistakes, and the honest counterpoint

• Renaming the PMO. The PMO becomes the APMO, the status report gets a Kanban board, and decisions keep being made the same way. The test is simple: what changed in who decides and in how money is allocated?
• Leaving Finance out. Without Finance in the group, the annual project budget keeps running underneath, and LPM becomes one more layer on top of it.
• Turning guardrails into bureaucracy. Guardrails exist to grant autonomy within limits. If every guardrail becomes a form and an approval queue, the result is the opposite of what was intended.
• Treating WSJF as mathematical truth. It structures the conversation about priority; it doesn’t replace the judgment of people who know the business.
• Never saying no. A funnel without no-go decisions is just a wish list. A healthy portfolio stops epics often, and that’s a sign of learning.
• Wanting everything at once. Implementing all eight steps perfectly before running the first cycle delays learning. An imperfect cycle today beats a perfect design six months from now.

And a few counterpoints that need to be said:

• Not every portfolio needs all of this. A company with three teams and one product doesn’t need Participatory Budgeting. The roadmap is for when several value streams compete for the same money.
• Regulated industries have real constraints. Banks, insurers, and healthcare companies have audit, regulatory, and accounting requirements that don’t go away. The way forward is to bring those areas into the group and design the guardrails with them, not to pretend the rules don’t exist.
• Changing the budget takes time. It’s common for the first year to run the old and new models side by side. That’s not failure; it’s transition.

AI in LPM: the group stops gathering data and starts deciding

A portfolio produces more information than any group can read: hundreds of Jira items, meeting notes, cost spreadsheets, customer research, market reports, and regulations. Reading, cross-referencing, and summarizing that volume is exactly the kind of work where AI is already faster than any person. An LPM group that doesn’t use AI today decides with less information than it could have, and more slowly than the market around it.

Three uses that change the game

1. An inventory in hours, not weeks (step 4). Connect AI to the sources you already have, through connectors (MCP) to Jira, Azure DevOps, spreadsheets, and documents, or through exports. Ask it to cross-reference everything and flag duplicate initiatives, initiatives with no strategic theme, and teams spread across too many fronts. What used to take weeks of spreadsheets becomes a draft ready for the group to decide on.

2. A skeptic for every epic (step 5). Before the go / no-go, AI reads the hypothesis and the Lean Business Case in the role of someone who will challenge the investment: which assumptions hold everything up, which indicators show up too late, what an MVP at half the cost would look like, which regulatory and data risks were left out. The Epic Owner arrives at the Portfolio Sync with answers, and the meeting decides instead of discovering.

3. Forecasts and scenarios instead of guesswork (steps 6 and 8). With the teams’ delivery history, AI runs a Monte Carlo simulation and answers questions like “what are the odds this epic finishes by June?” In Participatory Budgeting, it simulates scenarios: if 15% of the budget moves from one value stream to another, what happens to the key results and to the epics in the queue?

Agents: a portfolio that tracks itself

The next step is to leave AI agents running on a schedule, without anyone having to ask:

• Every Monday, an agent reads the Portfolio Kanban and alerts the group to epics stuck for more than 30 days and states over their WIP limit.
• Every month, another compares planned and actual budget by value stream and warns when a guardrail is close to its limit.
• Before each Portfolio Sync, an agent builds the agenda with what needs a decision. After the event, it records decisions, owners, and deadlines, and drafts the communication to the teams.
• Before the Strategic Portfolio Review, an agent summarizes what changed in the market, the competition, and industry regulation during the quarter.

The gain isn’t just time. The group starts seeing problems when they appear, not a month later.

Three prompts to use tomorrow

Inventory:

You are a portfolio analyst. I’ve attached the Jira export with all open initiatives, the cost spreadsheet per team, and the list of strategic themes. Build a table with initiative, owner, estimated monthly cost, linked strategic theme (or “none”), and teams involved. Then flag initiatives with no theme, possible duplicates, and teams on more than three initiatives at once. Don’t make decisions: list the facts and the open questions.

Epic skeptic:

Act as a skeptical member of the investment committee. Read the attached epic hypothesis statement and Lean Business Case. Point out the three weakest assumptions, the indicators that will only show up too late, what an MVP at half the cost would look like, and the regulatory and data risks that weren’t addressed. Finish with five questions the Epic Owner must answer before the go / no-go.

Forecast:

I’ve attached the weekly history of items completed by the teams in this value stream over the last 26 weeks and the number of remaining items in each epic. Run a Monte Carlo simulation with 10,000 iterations and report, for each epic, the date with a 50% and an 85% chance of completion. Show the code you used.

Asking for the code isn’t a detail: it’s what makes the forecast auditable and lets someone on the team check the calculation.

What still belongs to the group

Using AI forcefully doesn’t mean using it without judgment:

• The decision is human. AI shows the data and the scenarios. Whoever decides to invest or stop, and answers for it, is the group.
• Numbers behind a decision get checked at the source. Every number supporting an investment is verified before it reaches the table.
• Hard conversations aren’t delegated. Explaining to the sponsor of a stopped initiative why it was stopped is still leadership’s job.

Tools: start with what you already have

The whole first cycle runs on tools almost every company already uses. A dedicated platform makes sense later, once the group knows exactly what it needs.

To get started:

• Spreadsheets (Excel or Google Sheets): team costs, inventory, monthly budget, and guardrails. The portfolio cost spreadsheet covers these steps.
• Work board (Jira, Azure DevOps, or similar): the Portfolio Kanban, with the funnel states, WIP limits, and epics linked to the teams’ work.
• Digital whiteboard (Miro, Mural, or similar): Portfolio Canvas, SWOT, WSJF scoring, and the Participatory Budgeting groups.
• Wiki and documents (Confluence, SharePoint, Notion): LPM charter, hypotheses, Lean Business Cases, and the decision log.

To measure and forecast:

• BI (Power BI, Looker, or similar): the three-layer dashboard, pulling data from the work board and from Finance.
• Monte Carlo forecasting: tools such as ActionableAgile or the free Focused Objective spreadsheets. AI can do it too, with the forecast prompt from the previous section.

When the portfolio grows: portfolio management platforms such as Jira Align with Atlassian Focus, Planview, ServiceNow SPM, IBM Apptio Targetprocess, and Bizzdesign. Planview, ServiceNow, and Bizzdesign were named leaders in Forrester’s 2026 evaluation of strategic portfolio management tools.

AI across every layer: any assistant that reads files, runs calculations, and connects to the company’s sources. It’s what does the inventory, the epic analysis, the forecasts, the agendas, and the agents described above.

One caveat: tools don’t fix process. It’s common to buy the platform before defining who decides, and end up with the old PMO inside expensive software. Run the first cycle with a spreadsheet and a board, and choose the platform based on what you missed.

Where to start next week

If you can only do one thing, do step 4: list everything in progress and take one question to the leadership group: “what are we going to stop?” No tool, framework, or certification replaces the effect of reducing work in parallel.

The portfolio cost spreadsheet helps with that first step. It’s available on GitHub, along with every illustration in this article.

LPM isn’t about boards, events, or spreadsheets. It’s about leadership deciding together, with data, where to put the company’s money and attention, and having the courage to change that decision when learning calls for it.

So, in your company: who decides today that an initiative starts, and who decides that it stops?

Further reading

• SAFe: Lean Budget Guardrails
• SAFe: Portfolio Kanban
• SAFe: Epic Owners
• SAFe: Business Owners
• Beyond Budgeting Round Table
• Klaus Leopold, Rethinking Agile: Why Agile Teams Have Nothing To Do With Business Agility, the book that introduces the Flight Levels model
• Forrester: Why Strategic Portfolio Management Is Entering A New Era