The Planning Stack: Vision to Quarter Without Repeating Yourself

I once sat through a planning offsite where the leadership team presented the 5-year vision, the 3-year plan, and the annual plan back to back, and somewhere in hour two I started playing a private game: cover the slide header and guess which deck this slide belongs to. I couldn’t. The 5-year plan said “become the leading platform for X.” The 3-year plan said “establish market leadership in X.” The annual plan said “grow X revenue 40%.” Same nouns, same charts, different fonts on the ambition. When someone asked what we’d stop doing if the 3-year bet was real, the room went quiet, and then someone changed the slide.

That deck-matching game is the fastest planning audit I know, and it exposes the core failure: if your plans at different horizons say the same thing at different zoom levels, no planning happened at the longer horizons. The annual plan that’s just the quarterly plan times four means nobody made an annual decision. The 3-year plan that’s the annual plan with bigger numbers means nobody thought about what the company needs to become, only what it needs to do more of. Repetition across horizons isn’t alignment — it’s the absence of the distinct decision each horizon exists to make.

The fix isn’t better slides. It’s a rule: each horizon owns a question the others can’t answer. If a document doesn’t answer a question that no other document answers, it shouldn’t exist. This post is the stack I use — what each layer decides, what its one-page artifact looks like, who writes and approves it, and the smell test that reveals when it’s fake. It sits on top of the strategy work I’ve covered before — what makes a strategy real and how it cascades into a sequence — so I won’t re-argue any of that here. This is specifically about the horizons and their cadence.

The stack at a glance

Horizon	The question only it answers	Length	Written by	Approved by	Reviewed
Vision (undated)	What world are we trying to cause?	1 page	Founder/CEO	Board, once	Almost never
5-year	Which market position do we hold at maturity?	1–2 pages	CEO + exec team	Board	Yearly, rewritten rarely
3-year	What must be true of the company to reach that position?	2–3 pages	Exec team, CPO/CTO heavy	CEO	Yearly
Yearly	Where does the money and headcount actually go?	2–3 pages	Function leads, CEO edits	CEO + board	Quarterly check
Quarterly	What specific outcomes ship, owned by whom — and what don’t we do?	1–2 pages	Team leads	Exec sponsor	Weekly against reality

Notice what the table implies: the documents get more concrete going down, but they don’t get shorter versions of each other. The 5-year plan contains no features. The yearly plan contains no vision language. Each one is written in a different vocabulary because it’s deciding a different thing.

Vision: the customer’s life, not the company’s

The vision is undated and it describes the world, not you. The single edit that fixes most bad visions: write it as a description of the customer’s life after you’ve won, with your company nowhere in the sentence. “Every construction project manager knows their real cost position daily, not at month-end” is a vision. “Be the leading construction cost platform” is a wish wearing a vision’s clothes — I’ve covered why wishes aren’t strategy already, and the same disease infects visions.

One page, prose, no charts. The founder writes it because it’s the one document that’s allowed to be personal conviction rather than analysis. It changes approximately never — if you’re revising the vision more than once every several years, either the first one was wrong or you’re using “vision refresh” as a way to avoid harder conversations lower in the stack.

Smell test: read it to a customer. If they can’t recognize their own life in it — or worse, if they ask what your product does, because the vision only describes your product — it’s a horoscope: vague enough that any future confirms it, which means it constrains nothing.

5-year: which position, in which market structure

The 5-year plan answers where we play and how we win at maturity — and the key discipline is that it bets on market structure, not on features. Nothing you’re building today survives five years intact, so a 5-year plan full of product ideas is fiction. What it can honestly contain: which segment you’ll own, which alternatives you’ll have displaced, what the buying motion looks like, what your durable advantage is by then, and — the part everyone skips — what you believe the market will have done: consolidation, platform shifts, regulation, who commoditizes.

This is where positioning gets its long-run answer. One to two pages, written by the CEO with the exec team, approved by the board. Reviewed yearly — reviewed, not rewritten. The review question is narrow: “which of our market-structure beliefs did this year’s evidence confirm or break?” If none broke, the plan stands. Rewriting the 5-year annually is a tell that it was never a bet, just a mood.

Smell test: swap your logo for your nearest competitor’s. If nothing breaks — if they could present your 5-year plan without edits — it isn’t a plan, it’s an industry forecast. A real position is one your specific competitor could not adopt without abandoning something they’re committed to.

3-year: the capability plan

This is the layer most companies simply don’t have, which is why their 3-year deck is the annual deck with bigger numbers. The 3-year plan answers: what must be true of the company — not the product — for the 5-year position to be reachable? It’s written about capabilities: what the platform must support, what the team must know how to do, what the go-to-market motion must be, what the moat must consist of.

This is where the decisions live that are too slow for a yearly plan and too concrete for a 5-year one. Build-vs-buy on infrastructure. The platform investment that pays nothing for eighteen months. The shift from founder-led sales to a real motion. The second product line’s foundations. Opening an org capability you don’t have — data, hardware, enterprise security — which takes two years of hiring and learning before it produces anything. None of these fit in a yearly plan, because a yearly plan would always defer them; all of them are too specific for the 5-year page.

Two to three pages, exec-team authored with the CPO and CTO carrying most of the pen, approved by the CEO, reviewed yearly. The useful format is a short list of capability gaps: “today we can X; by year three we must be able to Y; the bridge is Z.” Five gaps maximum. If you have twelve, you haven’t chosen.

Smell test: does it contain at least one investment that makes this year’s numbers worse? A capability plan that’s all upside is just the revenue plan restated. Real capability bets cost you now to be possible later — if nothing in the 3-year plan hurts the yearly plan, one of them is lying.

Yearly: the only layer where money moves

The yearly plan is resource allocation, full stop. Themes, headcount, budget, and the two-to-four big rocks — and it’s the only horizon at which money is actually committed, which makes it the layer where the stack gets real or gets exposed. Everything above it is belief; this is where belief buys things.

Function leads draft their pieces, the CEO edits it into one coherent document — edits, meaning cuts, because the draft always contains six big rocks and the plan may contain at most four — and the board approves it alongside the budget. Two to three pages: the year’s themes (each traceable to a 3-year capability gap or a 5-year position claim), the headcount plan by team, and the big rocks with a sentence each on what “done” means. The roadmap is not the yearly plan — the roadmap is a downstream artifact built inside the yearly plan’s constraints, and the metrics per layer are their own discipline I won’t reopen here.

Smell test: can you point to something you funded less because of it? An allocation plan that allocated nothing away from anything isn’t a plan, it’s a payroll report. The second test: multiply the quarterly plan by four. If you get the yearly plan exactly, the yearly layer is empty — a real yearly plan contains at least one thing (a capability rock, a hiring bet, a platform investment) that no single quarter would ever choose for itself.

Quarterly: commitments and the not-doing list

The quarterly plan is execution: specific outcomes, named owners, and — the section that does most of the work — the explicit list of what we are not doing this quarter. The not-doing list is where the quarterly plan earns its existence, because the outcomes list is usually uncontroversial; it’s the deferrals that surface the disagreements the planning meeting exists to resolve. A quarterly plan without a not-doing list hasn’t decided anything, it’s just predicted its own activity.

One to two pages, written by the team leads who’ll be held to it, approved by an exec sponsor, checked weekly against reality. I’m deliberately staying shallow here — the next post in this series is entirely about running the quarterly layer, so consider this the doorway, not the room.

Smell test: is anything on the not-doing list something a specific stakeholder actively wanted? A not-doing list of things nobody asked for is decoration.

How the layers connect: constraints down, evidence up

The connective tissue matters more than the documents. Downward: each layer is a constraint on the one below it, not a summary of it. The 5-year position constrains which capabilities the 3-year plan may choose; the capability gaps constrain which themes the yearly plan may fund; the yearly themes constrain what the quarter may commit to. A layer that merely summarizes the one below it — the annual plan as the quarters stapled together — has the causality backwards: the bottom is writing the top, which means the top doesn’t exist.

Upward, the flow is evidence, and it needs an explicit escalation rule, because the default in most companies is that inconvenient quarterly evidence gets filed as “learnings” and the yearly plan sails on. The rule I use: quarterly evidence is allowed to falsify the yearly theme — but only at the quarterly review, in writing, with the claim stated as “this theme’s underlying assumption was X; here is the evidence X is false.” Not “the theme feels wrong,” not a hallway lobbying campaign. If the case holds, the yearly plan gets a mid-year amendment, logged as such. If yearly evidence breaks a 3-year capability bet, same mechanism at the annual review, one level up. Each layer can be reopened only by evidence from the layer directly below it, through its own review cadence. That single rule is what lets you have stable long-horizon plans and honest short-horizon feedback without the two constantly trampling each other — the same up-and-down causality I described in the cascade, here given a calendar and a paper trail.

One ambition, five altitudes

Here’s a single ambition — a field-operations SaaS wanting to own the mid-market — threaded through the stack, so you can hear how differently it reads at each level:

• Vision: “A site supervisor never re-enters data a machine already captured.”
• 5-year: “We are the system of record for field-to-office data in mid-market civil construction; ERPs integrate with us, not the reverse; the buying decision has moved from IT to operations.”
• 3-year: “To be integratable-with rather than integrating, we need a public API platform, a partner certification motion, and two enterprise-grade compliance certifications we don’t have today. Gap: we currently build every integration ourselves.”
• Yearly: “Big rock 2 of 3: the API platform. Two teams, one senior platform hire, at the cost of slowing the reporting roadmap that sales wants — accepted, board-approved.”
• Quarterly: “Ship API v1 to five design partners; owner: Asmita; success: two partners push production data. Not doing: the ERP-side connector Sales asked for — it’s the old model.”

Ten lines, one ambition, five different decisions. The vision never mentions the company. The 5-year never mentions the API. The 3-year names the gap. The yearly pays for it and names what it displaced. The quarterly ships a slice and refuses a request. Cover the headers on those five and you’d never confuse them — that’s the whole test.

Failure modes I keep meeting

The horoscope vision. Vague enough that any outcome confirms it, so it constrains nothing and the real top of the stack becomes the yearly revenue target — which is how companies end up strategic about nothing and aggressive about everything.

The 5-year plan as investor theater. Written for the data room, full of TAM slides, never consulted between fundraises. The tell: nobody below the exec team can state its central bet. The investor-facing view should be derived from the real plan, the way the external roadmap is derived from the internal one — the moment the fundraising artifact is the plan, you’ve optimized the document for people who don’t have to execute it.

The missing 3-year layer. Covered above, but worth naming as the most common structural hole: without a capability plan, every long-payback investment loses the annual budget fight to something with in-year revenue, three years running, and then the 5-year position is quietly unreachable and nobody made that decision on purpose.

Planning calendar capture. The stack described here should cost roughly: a day a year on the top two layers, two days on the 3-year, a week of elapsed calendar on the yearly, and a day per quarter. Companies that spend all of Q4 planning aren’t planning more, they’re re-litigating the upper layers every year because those layers were never actually decided — the whole point of a stable stack is that most horizons are closed most of the time, and the quarters get to just execute.

Put it to work

1. Play the deck-matching game. Pull your company’s vision, long-range plan, annual plan, and current quarter. Cover the headers, shuffle ten slides, and have someone assign each slide to a horizon. Every slide that could belong to two horizons marks a layer that isn’t doing its own job.
2. Run the five smell tests from this post against the corresponding document. Any document that fails, don’t rewrite it yet — first write down, in one sentence, the question that horizon should be answering for your company. The rewrite is easy once the question is honest.
3. Install the escalation rule before the next quarterly review: falsification claims in writing, assumption named, evidence attached, amendments logged. It takes one paragraph in the planning doc and it’s the difference between a stack and a shrine.

The next post in this series goes one level down and stays there: how to actually run quarterly planning — the meeting, the inputs, the commitment format, and how to keep two days of planning from consuming two weeks. The quarterly layer is where the whole stack touches the ground, and it deserves more than the doorway I gave it here.

Further reading

• Richard Rumelt, Good Strategy Bad Strategy — the diagnosis discipline that keeps the upper layers honest; I’ve written about applying the kernel already.
• Geoffrey Moore, Zone to Win — the sharpest treatment I know of why long-payback capability bets lose annual budget fights, and the governance that protects them.
• A.G. Lafley & Roger Martin, Playing to Win — the where-to-play/how-to-win vocabulary the 5-year layer is written in.
• Melissa Perri, Escaping the Build Trap — on the difference between a company that plans outcomes at every horizon and one that plans output at five zoom levels.