The Workshop: The Planning Onion

Five planning layers, from a daily standup out to the yearly vision. The Planning Onion is the workshop that connects them, every sprint goal traces back to a quarterly theme, and every quarterly theme traces back to a strategic choice. Worked example: From Weekly Habits to Yearly Vision.

The Planning Onion

The Planning Onion makes the planning layers visible at once, strategy, portfolio, quarter, sprint, day, and walks them outside-in and then inside-out in ninety minutes, producing a stated commitment, named owners, and a review cadence per layer, plus a list of the danglers (initiatives living on one layer with nothing above or below to anchor them). Sometimes called planning horizons, cascading objectives, or layered planning; sometimes associated with OKRs, though it predates them and works without them. The modern form traces through Stephen Bungay’s The Art of Action (2011) and the agile planning onion canonised by Mike Cohn in Agile Estimating and Planning (2005). Frequently confused with roadmapping: a roadmap is a time axis with work laid against it, the onion is a cadence structure with work nested inside it, and a roadmap can sit on any one layer.

At a glance

Who, for how long: a facilitator who doesn’t carry a layer, delivery lead(s), team leads from squads in scope, a product lead, and optionally a founder or exec for the outer rings. Five to nine people, ninety minutes.
What you walk out with: a filled onion with traces drawn between layers, a vertical-slice result naming where the trace breaks, a categorised dangler list (kill / missing above / missing below / cross-cutting), and a review cadence per layer with owners and first dates.
When to reach for it: a felt sense that sprint work, quarter plan, and strategy don’t match, or a new strategy that you want to check against execution. Not for writing strategy from scratch (run Business Model Canvas first), sprint-only conversations (use Sprint Planning instead), or rooms where leadership strategy is off-limits.

What’s It For

A quarterly initiative called “improve reliability” has been on the plan for two months. Somebody is working on it. It produces no measurable output week to week, because there’s no sprint work associated; the engineer chipping away at it doesn’t have a current story to point at. At the same time, on the daily stand-up, a team is completing sprint items that don’t appear anywhere on the quarterly plan because they were urgent when they started and nobody’s mapped them back up. At the top, a strategy document says “we’re the trusted daily habit for busy families.” Nobody on the team can name a sprint item this month that serves that strategy.

Three layers, three disconnects, no obvious villain. The Planning Onion calls the effect out: every layer is doing its own work, but the layers aren’t connected in the way the shape promises. Initiatives live on one layer with nothing above or below. Work happens that can’t be traced. Strategies are stated that don’t show up anywhere in execution.

This session exists to force the picture into the same room at the same time. The onion’s layers are easy to describe abstractly; they resist being kept honest without periodic inspection. The workshop is that inspection: fill each layer, walk outside-in, trace a sprint item back to strategy, find the danglers, commit to the cadences that keep the layers in sync.

Absent the session, each layer’s planning happens on its own schedule and with its own stakeholders. The quarter gets replanned in July, the sprint gets planned every two weeks, the day gets planned in stand-up. The strategy gets revisited once a year. The layers drift.

Reach for it when:

The team has a felt sense that their sprint work and their quarter plan and their strategy don’t match, but nobody’s mapped it
A new strategy has landed and you want to check whether it’s showing up in execution yet
Several initiatives have been running for weeks without visible output and the team isn’t sure whether they’re still live
You’re about to plan a new quarter and want to sanity-check that the inputs from strategy and portfolio have been received
Two teams are planning against different mental models of how the layers connect

What It’s Not For

Skip it when:

The strategy doesn’t exist. You’ll spend the session writing one, which is a different workshop; run Business Model Canvas first.
Only one layer is in scope. If the conversation is purely at the sprint level, use Sprint Planning instead.
The team isn’t authorised to see the outer layers. Leadership strategy excluded from the room means the outer layer is fiction, and the session will be theatre.
The review cadences are the real problem and you know it. Fix the cadences directly; don’t layer a workshop over a scheduling problem.

Stop signals once the session is running:

Strategy is fiction and can’t be fixed in this room
The team can’t name shipped work from the last three sprints
The layers collapse into one or two, there isn’t an onion, there’s a plan

Stopping and naming that the onion doesn’t exist yet is not failure. Producing a coherent-looking trace across layers that the team doesn’t believe in is.

Trade-offs to weigh going in.

Benefits:

A single view of the planning cascade that the team can point at for months
A list of danglers (initiatives that cost capacity without being connected to anything) with a kill-or-anchor decision per dangler
A vertical-slice trace that surfaces which layer is most broken
Review cadences committed, with owners and dates, at each layer
Strategy’s presence (or absence) in execution made explicit rather than assumed
A defensible answer to “why are we doing this?” for each piece of work

Costs:

5-9 people times 90 minutes, plus a pre-session data-gathering pass
Political cost when danglers are pet initiatives
Emotional cost when the session reveals strategy doesn’t land in execution at all
Recurring cost: the onion wants inspecting quarterly at minimum; one session doesn’t fix the cascade

Failure modes:

The onion is filled, the danglers are named, and nothing is killed or anchored, so they re-appear next quarter as the same danglers
The cadences get committed but never happen, and the layers silently drift again
Strategy-layer debate consumes time meant for the inner layers, and the vertical slice never runs
The vertical slice only runs on known-good items and produces false confidence
The session becomes a theatre in which the team rehearses a coherent story they don’t live in

Definitions & Background

Canonical five layers, inside to outside:

Day / week. What the team is doing right now. Stand-ups, daily practices, this week’s committed sprint items. Cadence: daily and weekly. Owner: the team. Commitment: the work completes this week.

Sprint. What the team will deliver in the next iteration, usually two weeks. Cadence: every two weeks. Owner: the team with the delivery lead. Commitment: a shippable increment.

Quarter. What the team will move by quarter-end. Cadence: per quarter, reviewed monthly. Owner: the delivery lead and product lead jointly. Commitment: outcomes and milestones for the next three months.

Portfolio. Which initiatives exist across the team or teams, and how they compete for capacity. Cadence: quarterly, reviewed monthly. Owner: the product lead. Commitment: a set of live initiatives, each with a stated quarter’s goal.

Strategy. The why. What kind of business we’re building, for whom, against what constraints. Cadence: annually, reviewed quarterly. Owner: the founders or execs. Commitment: a set of directional choices that make everything below coherent.

A note on Year vs Strategy. Some teams running Vision / Year / Quarter / Sprint / Day rhythms split the outermost layer in two: Vision (the long-term why; changes rarely) and Year (annual themes or bets, refreshed each January). This playbook collapses them into one because most teams find Year and Vision get reviewed in the same conversation. If you run a deliberate annual planning cycle that’s distinct from a less-frequent Vision conversation, treat them as two layers and run the workshop with six rings instead of five.

Each layer commits to a horizon and a cadence. The layer inside is constrained by the layer outside: sprint items are shaped by quarter commitments, quarter commitments are shaped by portfolio, portfolio by strategy. Violations of that constraint are the session’s primary finding.

Inputs

Layers in scope, named before the session starts. The facilitator and a sponsor agree which rings are live: all five, the inner three, or some combination. A scope written on the wall before the room arrives prevents the session opening with a definitional argument.
A rough document per layer: the current strategy (if the session covers it), the current portfolio commitments, the current quarter plan, the current sprint backlog, the team’s standing daily practices. These don’t have to be polished; they have to exist.
Wall space or a board big enough for five concentric rings, sticky notes in at least two colours (one for layer content, one for danglers), markers, and cards for the cadence commitments.
A 90-minute slot with no interruptions and the right people in the room (see Who’s Needed).

Outputs

What lands on the wall at the end:

A filled onion with each layer in scope populated (strategy, portfolio, quarter, sprint, day) in the team’s own language.
Traced lines between layers, drawn during the outside-in walk, showing which sprint items descend from which quarter commitments, which quarter commitments descend from which portfolio initiatives, and so on up to strategy.
A vertical-slice result: the random sprint item walked upward, with the layer (if any) where the trace broke, named explicitly.
A categorised dangler list: each dangler tagged as should be killed, missing the layer above, missing the layer below, or legitimately cross-cutting.
A review cadence per layer, written on cards, each with a named owner and a date for the first instance.
A short summary message: here are the layers, here’s what’s connected, here are the danglers, here’s when each layer gets revisited.

Photograph the onion with its lines drawn and its danglers flagged before the wall comes down.

These outputs feed straight into:

Impact Mapping. Impact Mapping produces the quarter-layer content; the Planning Onion checks whether the quarter connects to strategy and down to sprints. Impact Mapping is input; the Onion is inspection.
Business Model Canvas, the Canvas shapes the strategy layer. When the Onion reveals strategy is missing, the Canvas is the session to run before the next Onion pass.
Prioritisation, within any layer, prioritisation orders the work. The Onion reveals which layer needs prioritisation first, often the quarter layer, where commitments are stated but competing.
Sprint Planning, sprint planning is the inner-layer cadence the Onion commits to. If sprint planning never lands, the whole onion drifts upward.
Wardley Mapping. Wardley tells you how the landscape is evolving; the Onion tells you whether your plan descends coherently through it. Composes well for strategy re-writes.
Cynefin Sensemaking, some quarter commitments are complex rather than complicated. When the Onion reveals a layer-below the team can’t fill, Cynefin often explains why: the work is complex and wants probes rather than sprint items.
User Story Mapping. Story Mapping lays out the journey at the sprint-to-quarter seam. The Onion points at that seam; Story Mapping does the work there.

Who’s Needed

Five to nine people, ninety minutes:

Facilitator. Runs the clock, holds the layer discipline, and protects the vertical-slicing phase from sliding back into layer-internal debate.
Delivery lead(s). Mandatory. One per squad in scope. They own the mapping between layers for their work.
Team leads from squads in scope. The people carrying the sprint and day layers. Without them, the inner layers are a guess.
Product lead. Carries the quarter and portfolio layers. Often the person who will discover that their quarter plan contains two initiatives nobody on the team has capacity for.
Founder or exec (optional, briefly). If the strategy layer is in scope and evolving, a founder/exec attends for the strategy-and-portfolio phases and leaves before the sprint/day phases. They should not stay for the danglers discussion, which changes the honesty of the conversation.
Optional outside observer. Someone from an adjacent team who attends silently and asks one question at the end. Fresh eyes catch the danglers the team has stopped noticing.

Below five, the layers don’t all have someone carrying them; above nine, the vertical-slicing test becomes too slow to reach before the clock runs out.

Who to leave out:

Stakeholders without authority over any layer. The session’s output requires commitment at each layer; observers without commitment absorb oxygen.
Large groups of engineers. One or two per squad is enough; the rest get the output.
Anyone who wants to re-open the strategy during sprint discussion. Strategy debate has its place; it’s not here.

How To Run It

The session scales by the number of layers in scope.

Scope	Duration	When
Three layers (quarter, sprint, day)	60 min	Team-internal session; strategy and portfolio fixed elsewhere
Five layers (default)	90 min	Whole onion, with an exec or founder briefly in the outer layers
Full onion with multiple squads	Half day	Coordination session across teams; rarely needed

Default to five. Three works for tactical team-internal inspection; a half-day is almost always too long unless multiple squads are being coordinated.

Phase	Duration	Materials	Key question
Frame the onion and agree layers in scope	10 min	Layer diagram	“Which layers are we looking at today?”
Fill each layer individually	20 min	Sticky notes per layer	“What lives at this layer right now?”
Walk the layers outside-in	20 min	Filled layers	“Does each layer descend from the one outside?”
Vertical slice test	15 min	One random sprint item	“Can we trace this up to strategy?”
Find the danglers	15 min	Dangler list	“What lives on one layer with no anchor?”
Commit to review cadence	10 min	Cadence card per layer	“When does each layer get revisited?”
Total	~90 minutes

The vertical-slice test is the session’s most informative phase. If it goes wrong early, shorten the dangler phase and spend the time on the slice. One failed trace is worth five minutes of listing initiatives.

Phase 1. Frame the onion and agree layers in scope (10 min)

Draw the onion on the wall: five concentric circles, labelled from the outside in. Walk the room through the canonical layers, one sentence each.

“Five layers. Strategy at the outside: annual, owned by the founders. Portfolio inside that: the initiatives we’re running this year. Quarter inside that: what we’ll move by September. Sprint inside that: what we’ll ship by the end of the fortnight. Day-and-week at the centre, the stuff we’re doing right now. Today we’re looking at [all five / the inner three].”

Name the layers in scope explicitly. If strategy is not in scope, write “fixed” next to the outer ring and move on. The session can still produce useful findings at the inner layers, but the facilitator has to hold the outer ring as given. Arguments about strategy get parked.

What to watch for:

Layer confusion. The team calls two layers by the same name (“the quarterly sprint”). Fix it now; the session depends on the layers being distinguishable.
Missing layers. “We don’t have a portfolio layer.” Often true in small teams, where portfolio is collapsed into quarter. Name the collapse: “In our case, portfolio and quarter are one layer.” The onion still works with four layers.
Imported layer names. The team uses OKR language (objectives, key results) or SAFe (Scaled Agile Framework, a heavyweight enterprise-agile method) language (epics, features, stories, standard agile work-item sizes from biggest to smallest) at a specific layer. Keep their language but make sure the layer’s horizon and cadence are the working definitions, not the vendor’s.

Phase 2. Fill each layer individually (20 min)

Each layer gets five minutes. The participants whose role carries a layer write what they know about the current state onto notes and stick them in the right ring.

Strategy: what’s the current strategic direction? One or two sentences, or a short list of choices. What are we saying no to as well as yes to?
Portfolio: what are the current live initiatives? One note per initiative. Include the ones that have been running quietly in the background.
Quarter: what are this quarter’s commitments? By layer, by squad.
Sprint: what’s in the current sprint backlog, by squad?
Day: what are the team’s daily practices that consume capacity? Stand-ups, on-call rotations, review meetings, support shifts.

Silent writing first, then place on the layer. Fifteen minutes of individual work and placement; five minutes of tidying.

What to watch for:

Empty layers. An empty layer is a finding. Don’t fill it with speculation. Leave it empty and ask in Phase 3 why it’s empty.
Over-filled layers. Twenty notes on the sprint ring is normal. Twenty on the portfolio ring means portfolio is being used as a wish list, not a commitment.
Aspirational notes. Someone writes a strategy the room aspires to rather than the one the business actually has. Clarify: “Is this the strategy today, or the one we want?” Park aspirations in the dangler pile for Phase 5.
The repeated note. The same note appears on two layers, which is fine if the layers genuinely overlap, and suspicious if the note is trying to look busy.

Phase 3. Walk the layers outside-in (20 min)

The facilitator walks the outermost layer in scope, reads each note aloud, and asks:

“For each thing on this layer, what does the next layer in look like in service of it?”

From strategy to portfolio: which initiatives descend from which strategic choices? Some initiatives will trace cleanly to a strategy note; others won’t. Draw a line on the wall for each traced connection.

From portfolio to quarter: which quarter commitments descend from which portfolio initiatives? Again, lines where they exist.

From quarter to sprint: which sprint items descend from which quarter commitments?

From sprint to day: which daily practices support sprint completion, and which don’t?

Some lines will be short and clean. Some will not connect. The unconnected notes become a candidate dangler list: anything that doesn’t link to the layer outside it is a potential dangler, to be investigated in Phase 5.

What to watch for:

Confabulated connections. Someone draws a line between a sprint item and a quarter commitment that sounds plausible but doesn’t actually serve it. Push: “If this sprint item completes, which quarter commitment moves forward?” If the answer is “none specifically,” erase the line.
Strategy-to-everything. Every portfolio initiative gets traced to the same one-line strategy. Often the strategy is too abstract to constrain anything. Mark it for the danglers phase.
Missing descents. A layer has commitments that have no presence on the layer inside. That’s a dangler in the making; note it.
Work without a home. A sprint item nobody can connect upward. Possibly legitimate emergency work; possibly a dangler that’s been running on momentum.

Phase 4. Vertical slice test (15 min)

Ask the room to pick a sprint item at random. Genuinely at random, not a known success story. Often the best method is “third item from the top of squad X’s board.” Then walk it upward:

“This sprint item: which sprint commitment does it serve? Which quarter commitment does that sprint commitment descend from? Which portfolio initiative does the quarter commitment serve? Which strategic choice does the portfolio initiative belong to?”

Walk upward explicitly, layer by layer, drawing the vertical line on the wall. If the chain holds, the onion is doing its job for this item. If it breaks, the trace can’t be completed, the connection too weak to articulate, you’ve found something.

Pick a second item. Then a third if time allows. A single break is informative. Three breaks in three random items is a systemic finding.

What to watch for:

The broken chain. The trace fails at a specific layer. That’s the layer most in need of attention. Note which layer breaks most often.
The fluent-but-hollow chain. Someone narrates a coherent trace that everyone in the room silently disbelieves. Ask the person who’s not speaking: “Does that trace match your understanding?”
The trace that loops. The sprint item is supposedly serving a quarter commitment that’s supposedly serving itself at a different layer. Untangle and name: the layers aren’t actually different.
The retro-traced item. The trace only works because it was invented on the spot. That’s useful information: the team is improvising the onion rather than living in it.

Phase 5. Find the danglers (15 min)

Now catalogue the unconnected notes from Phases 3 and 4. Each dangler is a note on a layer that has no connection to the layer outside it, or no presence on the layer inside it, or both.

For each dangler, one of four things is true:

It shouldn’t exist. The initiative should be killed. Name it, mark it, propose the kill.
The layer above is missing. There’s no strategic commitment behind it; either the strategy needs updating to include it, or the initiative doesn’t actually deserve resources.
The layer below is missing. The quarter commitment is stated but the team hasn’t planned any sprint work against it. Either schedule the work or kill the commitment.
It’s a legitimate cross-cutting concern. Some work (security, compliance, platform upkeep) legitimately lives across the onion differently from product work. Mark it as cross-cutting and accept that the standard trace doesn’t apply, but name the alternative accountability.

The room decides per dangler. The output is a list, categorised.

What to watch for:

The beloved dangler. A pet initiative nobody wants to kill. Name it: “This is a dangler. Either we give it a strategic home this session, or we stop doing it. Which?”
The cross-cutting excuse. Everything becomes cross-cutting. Use it sparingly; half the team’s work isn’t actually cross-cutting.
Danglers rescued by re-writing the layer above. Someone writes a new strategy note to justify an existing initiative. Sometimes legitimate; often procrastination. The test is whether the new strategy note would have been written even without the initiative needing it.
Silence around a dangler. Nobody wants to address one. That’s the one most worth addressing. Facilitator’s job: “Let’s talk about this one. What is it, and does it belong?”

Phase 6. Commit to review cadence (10 min)

For each layer in scope, name when it gets revisited, by whom.

Day layer: stand-up, daily.
Sprint layer: sprint planning and review, fortnightly.
Quarter layer: monthly check-in plus quarterly replan.
Portfolio layer: monthly.
Strategy layer: quarterly (review) and annually (revise).

Write each cadence on a card. Confirm who owns calling each one. A layer without an owned cadence is a layer that will drift into the dangler list next quarter.

“If we don’t revisit portfolio monthly, we’ll be back here in three months finding the same danglers. Who owns the monthly portfolio review? When’s the first one?”

End the session on commitments, with dates.

Worked example

See The Planning Onion: From Weekly Habits to Yearly Vision for the Greenbox team’s first planning onion session, including the moment the vertical slice reveals that a full engineer has been working on an initiative that doesn’t connect to any quarter commitment, and the different-shaped relief of naming it rather than defending it.

What Can Go Wrong

The strategy layer is full of verbs. “Improve reliability, grow subscribers, delight customers.” No commitment, no trade-off, no no. Recovery: “What’s a thing we’ve decided not to do because of this strategy?” If the answer is nothing, the strategy is a vision statement and not yet a strategy. Stop if: The room can’t produce a no. The session can continue at the inner layers, but flag the strategic drift explicitly.

Quarter-level theatre. The quarter commitments are phrased so nobody could plausibly fail, goals nobody would bet against. Recovery: “If we came back in three months and half of these weren’t delivered, would we be surprised?” If no surprise either way, the commitments aren’t commitments. Stop if: The theatre is the norm. Name it as a finding; the fix is a different session.

Invisible sprints. The sprint layer has items but the team can’t name what shipped in the last three sprints. Recovery: Ask for concrete shipped work by name. If none emerges, the sprint cadence is broken and the whole onion below quarter is hollow. Stop if: The team can’t produce shipped work from memory. You’ve found a bigger problem than planning.

Reviews that never happen. The cadence phase produces commitments the team knows won’t land. Recovery: “What prevents the monthly review from happening today? Let’s fix that constraint, not layer a commitment over it.” Stop if: The team can’t name what prevents the review. Something larger is wrong; the session won’t fix it.

Everything-is-a-dangler. The dangler phase fills with every initiative on the wall. Recovery: The trace from Phase 4 is a better guide than individual inspection. Prioritise the danglers that failed the vertical slice; ignore the rest until those are handled. Stop if: The team’s morale collapses. End with a short list of the most urgent danglers and reschedule the rest.

Senior stakeholder redefining layers on the fly. The exec in the room decides the layers are different from what the session started with. Recovery: Accept the redefinition if it’s substantive. Park it if it’s cosmetic. Stop if: The exec wants to re-run the session their way. Let them schedule it; today’s session completes at the inner layers as it stands.

Next Steps

The session ends; the work begins.

Same day, the facilitator:

Photographs the onion with its lines drawn and its danglers flagged.
Transcribes the list: each layer’s commitments, the traces between layers, the danglers with their category, the review cadences with owners.
Sends a short summary message: here are the layers, here’s what’s connected, here are the danglers, here’s when each layer gets revisited.

The delivery lead’s fortnight. This is where the pattern earns its cost.

Kill or anchor each dangler within a fortnight. A dangler allowed to persist beyond the session becomes a quiet contributor to the next quarter’s dangler list. The kill is the harder commitment; make it explicit, with a date.
Schedule the first review at each layer. The monthly portfolio review, the monthly quarter review, the quarterly strategy review. Put them on calendars immediately. An uncommitted cadence is a cadence that won’t happen.
Re-read the vertical slice in context. If the slice broke at the quarter-to-portfolio seam, that’s where the next structural conversation is. The fix isn’t necessarily the onion session; it might be a separate portfolio workshop.
Walk the findings past anyone who wasn’t in the room. Especially leadership if they weren’t present for the strategy phase. Their reaction reveals whether the strategy layer is shared or the team’s reconstruction of it.

Ongoing:

Pin the onion photograph somewhere the team works. When new work is proposed, ask which layer it lives on and what it’s connected to before accepting it.
Track the dangler rate. If more danglers appear each quarter, the cadences aren’t working, reviews are happening but not producing decisions.
Re-run the vertical slice in every quarter’s planning session. One random sprint item walked upward, every quarter, is a five-minute health check for the whole cascade.

Variants

Five-layer (default). Strategy / Portfolio / Quarter / Sprint / Day, ninety minutes, five to nine people. Output: filled onion, traces, vertical slice, dangler list, cadences. This is what most teams need, and the rest of this post describes it.

Three-layer (team-internal). Quarter / Sprint / Day, sixty minutes. Run when strategy and portfolio are fixed elsewhere and the team needs tactical inspection. Output is the same shape but stops at the quarter ring; the outside-in walk starts there and arguments about strategy are explicitly out of scope.

Six-ring (Vision / Year / Quarter / Sprint / Day plus a separate Strategy or Portfolio ring). Run when a deliberate annual planning cycle is distinct from a less-frequent Vision conversation. Vision changes rarely; Year is the annual themes or bets refreshed each January. Adds ten to fifteen minutes to the fill and outside-in phases; the vertical slice still runs once.

Full onion with multiple squads (scaled-up enterprise onion). Half day, multiple delivery leads and product leads, one onion per squad on adjacent walls, with a shared outermost strategy ring. The vertical slice runs per squad; the dangler phase looks for danglers that appear across squads (which usually indicates a missing portfolio layer rather than per-squad failures). Reach for it when coordinating two or more squads against the same strategy; rarely needed, easy to over-use.

Remote. Miro or Mural board with five concentric rings drawn, sticky notes per layer, video call for the conversation. The vertical-slice phase needs the facilitator to draw the upward trace live so the room sees the chain forming or breaking. Slightly slower than in-person; the structure transfers cleanly.