From a raw marketing plan to a decision document a board signs
For marketing leads and founders who have to defend a budget. Nine sections, five financial rules and an internal council that burns your plan down before the board does.
Most marketing plans are not decision documents. They describe what the team wants to do, with a budget that is "about right". Then the CFO asks where that amount comes from, and the plan slips a quarter.
This guide is for marketing leads and founders who have to put a plan in front of a CMO, a board or an investor, and who want it to survive its first critical reader.
What's inside
- The nine sections of a decision document, and why budget only comes in section seven
- Five financial rules, starting with: you derive a budget, you never assume one
- The block almost every plan misses: what the pilot can not prove
- The internal council: five reviewers with clashing lenses, and the one question that yields the biggest find
- A free skill for Claude that runs the whole pipeline, with template, validator and a fictional example plan
Where should we send the guide and the skill?
A pretty plan is not a decidable plan
A decision document has one job: enabling someone with budget authority to say yes, no or "this first". That takes three things most plans lack. Every euro is derived from unit economics. Every assumption is labeled or replaced by data. Every risk has an owner.
The method below runs in four phases: diagnosis, document, review by a council, rework. You can use each phase on its own.
Phase 1: the diagnosis
Read all the input and write at most ten bullets in four categories: what is missing, what is illogical or in the wrong order, what is weakly supported, and what is strong and stays. Concrete, with reference to the input. Put the diagnosis in front of whoever wrote the plan before you rewrite a single sentence. What you agree on in this step decides what you fill in and what you take over.
Phase 2: nine sections
The red thread: context, audience, objectives, strategy, execution, money, risk, decision. Budget comes after the strategy, never before. The reader has to understand why first, and only then how much.
| # | Section | What it has to do |
|---|---|---|
| 1 | Executive Summary | Situation, complication, resolution and a "decision required" block. Stands fully on its own. |
| 2 | Strategic context | Market with four key numbers and source, competitive field, and the market definition question as its own subsection |
| 3 | Audience and segmentation | Personas, the psychology of the decision moment, jobs to be done in the customer's words |
| 4 | Objectives and KPIs | One north star with a baseline, decision rules, and what the pilot can not prove |
| 5 | Go-to-market | One promise, messages per stage, channels with rationale |
| 6 | Roadmap | Phases with calendar realism, what needs building, a two-step test approach |
| 7 | Budget and resources | Unit economics, payback, sensitivity table, scenarios |
| 8 | Risks | Five columns, one of which is the owner |
| 9 | Next steps | Decision points with proposal and date, actions with role and deadline |
The market definition question deserves attention. Is your real competitor another player, or a substitute such as "included for free", "my supplier already does that" or "doing nothing"? In practice, a market definition that had never been tested was the biggest blind spot the council caught. Treat it as its own subsection with a validation plan, never as a footnote.
Five financial rules
- Budget = goal times blended CAC. You never take a budget as given. You derive it and propose it, with a sensitivity table of CAC levels against goals, so every combination can be discussed.
- Payback is a design criterion. I work with a payback per customer between one and two years. That immediately caps your CAC. Replace that value with what your own margin allows.
- A margin sensitivity table. Show how much retention your ratio requires at different contribution margins. Label every ratio as revenue-based until the real margin is in.
- Brake and accelerator, fixed in advance. Four decision rules: accelerate, scale, adjust with one retry, redesign. Tie every threshold to payback and ratio, not to round numbers. Then the decision after the pilot becomes mechanical.
- Capacity belongs in the scenario decision. A scenario that leans on warm channels gets calculated as a formula with a labeled example and the minimum staffing in FTE.
Never invent numbers. What you do not know is a labeled assumption or a question for whoever does have the number.
The block almost every plan misses
Write down what the pilot can not prove. Which metrics fall outside the pilot window, such as conversion after ninety days? Where are the numbers too small to decide on? Which assumptions can only be tested through interviews or partnerships? A pilot that promises more than it can measure produces false certainty. And a north star without a baseline is decoration.
Phase 3: the council
Whoever wrote the document is the worst placed reviewer of it. That is why it goes past an internal council first. The mechanism comes from Andrej Karpathy's LLM Council idea: independent answers, anonymized peer review, a chairman who synthesizes. My addition is the framing. The council does not judge an open question but a decision document, with the stakes spelled out: a weak document costs months or buys a wrongly scaled budget.
Five reviewers, started at the same time so nobody reads anybody else:
| Lens | Looks for |
|---|---|
| The fault finder | The fatal flaw, what is missing, what will fail |
| The first principles reader | Whether you are asking the right question at all |
| The upside hunter | What could be bigger, which opportunity is hidden |
| The outsider | Jargon, claims without source. Deliberately gets no background. |
| The executor | Owners, deadlines, capacity, what happens on Monday morning |
Then five reviewers judge the anonymized answers on three questions. The third matters most: what did all five miss? In practice the biggest find only surfaces there, because those reviewers see the shared frame of the first round from the outside.
The chairman delivers five blocks: where the council agrees, where it clashes, which blind spots it caught, the recommendation, and the one thing to do first. The verdict "do not send this to the decision maker yet" is a valid outcome.
Phase 4: the rework
Three iron rules:
- Nothing irreversible. Backup before the first change. Send or publish nothing.
- No invented numbers. What can be measured internally becomes a data request: every assumption for which the company has the real number in a database. What can be tested externally becomes research with source and verification status.
- The document ends more honest than it started. What is not proven is in there as a hypothesis.
The data request is often the most valuable by-product. It costs nothing and decides everything: churn per cohort, conversion between the tiers of your offer, contribution margin. Aggregated numbers are enough. If the research shows a core claim does not hold, change the positioning across the whole document. Do not hide the finding.
The quality bar
- Does the Executive Summary stand fully on its own?
- Is every budget derived, with a sensitivity table?
- Does every KPI have a measurement definition, and the north star a baseline?
- Does the pilot validate the largest block of the goal?
- Does every risk have an owner and every action a role plus deadline?
The skill
The Strategy Council runs the four phases, with a checkpoint after each. In the zip: the skill, the document architecture, the playbooks for the council and the rework, an HTML template that prints on A4, a validator, and the raw plan of a fictional client to practice on. Everything in English and Dutch.
Download the Strategy Council skill (zip)
The council uses parallel subagents and therefore works best in Claude Code: copy the folder to ~/.claude/skills/. In the Claude app the diagnosis and the document work fully. No Claude? The nine sections, the five rules and the quality bar above also work as a checklist next to your own plan.
One next step
Take your current plan and find the budget. Does it say how it was derived? If not, do the math: goal times blended CAC. The gap between that number and what is on the page is your first agenda item.