There is a specific test for whether a document is a marketing plan.
Hand it to someone who doesn't work in your business and ask them to spend next month's budget from it. If they can — if they know who to reach, where, with what, for how much, and how you'll know it worked — it's a plan. If they'd have to come back with questions, it's a description of intentions.
Most documents called marketing plans fail this test. They contain goals, a brand statement, a list of channels, and a calendar. All reasonable content. None of it decides anything.
Nine parts
1. Who, specifically
Not a demographic. Not a persona with a stock photograph and a name.
A useful audience definition names a situation: what the person is trying to do, what's currently in their way, and what triggers them to look for help. "Small business owners aged 30–50" is a filter. "Owners of five-to-twenty-person service businesses who are doing marketing themselves at 9pm and have just lost a deal to a competitor with a better website" is a target.
The second version tells you what to write. The first doesn't.
Most plans should name two or three of these, ranked. Beyond three, resources fragment and nothing reaches sufficient depth to work.
2. What they'd do instead
Every marketing plan competes with an alternative, and it is usually not a competitor. It is a freelancer, an internal hire, a spreadsheet, or continuing as they are.
Naming the real alternative changes the message. Content that argues you're better than a competitor is wasted on someone who was never going to buy from that category at all.
3. The claim
One sentence: what you do, for whom, and why it's different. It must be something a competitor could not honestly say.
The test is negation. If the opposite of your claim is absurd — nobody would ever claim to be slow, unreliable and expensive — the claim carries no information. "High-quality service" fails. "Cheaper than a consultant and faster than doing it yourself" survives, because a competitor could plausibly claim the reverse trade-off.
This sentence governs everything downstream. Written badly, every subsequent decision inherits the vagueness.
4. Channels, with reasons
Not a list of channels. A list of channels with a stated reason each one is included, and — the part usually missing — a note on what is being excluded and why.
A plan that includes every channel has made no decisions. A small business can sustain two, possibly three, at a quality that works. The choice of which two is the most consequential decision in the document, and it should be defended in writing.
The reason should refer to where the audience already is, not to what's popular.
5. Budget, allocated
A total is not an allocation. The plan should say how much goes to each channel, over what period, and what happens to the money if something isn't working after a defined checkpoint.
This last clause is what makes a plan survive contact with reality. Without a stopping rule, underperforming spend continues by default, because stopping requires a decision and continuing doesn't.
| Channel | Allocation | Period | Stop condition | Reallocation |
|---|---|---|---|---|
| Paid search | 40% of budget | 3 months | Cost per qualified enquiry above £X after 6 weeks | To content if stopped |
6. What gets made
The calendar most plans consist of belongs here — but it needs a preceding section that says what kind of thing gets made and why.
Twelve blog posts is a quantity, not a strategy. Twelve articles targeting the specific questions your buyers ask before they know your category exists is a strategy, and the calendar follows from it automatically.
7. Measurement, chosen in advance
The metric has to be picked before the campaign runs, because a metric chosen afterwards will always be one that looks good.
For most small businesses the honest primary metric is enquiries or sales. Impressions, reach and follower growth are diagnostics — useful for understanding why something did or didn't work, dangerous as goals.
Each channel should also have a leading indicator, since the real metric arrives too late to steer by. Organic search: rankings for target terms at week six. Paid: cost per qualified enquiry at week two.
8. A schedule
Dates, and who does each thing. If a plan doesn't say who, it says nobody.
9. The assumptions
The section almost nobody writes, and the one that makes a plan revisable.
List what has to be true for the plan to work: expected conversion rate, cost per click, how long content takes to rank, whether the audience definition is right.
When results diverge from the plan — and they will — this list is what turns confusion into diagnosis. Without it, a plan that underperforms produces a generalised sense that marketing isn't working, rather than a specific finding that one assumption was wrong.
What doesn't need to be in it
A mission statement. It doesn't change any decision.
A full SWOT. The useful content ends up in the audience and competitor sections anyway; the grid adds ceremony.
Every channel that exists. An inventory is not a decision.
Industry statistics from a report. They pad the document and inform nothing.
A brand personality section, unless someone else is writing the copy, in which case it belongs in a brief rather than the plan.
Length
Shorter than expected. Nine sections done properly is five to ten pages.
Long marketing plans are usually long because they contain research that should have been an appendix, or because length is standing in for confidence. The plan is the decisions. The research is why.
When to revisit
Quarterly for the whole document. Monthly for the measurement section, against the assumptions list.
Rewriting the plan every month is a sign the plan wasn't specific enough to be wrong.
Questions
How long should it take to write?
The research takes days. The plan itself, once the research exists, takes hours. If writing the plan is taking weeks, the research isn't finished.
Do I need one if I'm a solo business?
More than a larger business does, because there's nobody to compensate for an unclear direction. It can be shorter — three pages covering audience, claim, two channels, budget and metrics.
Can AI write a marketing plan?
It can draft structure and produce good first passes at sections once you supply the inputs — audience research, competitor findings, budget, constraints. What it cannot supply is the decisions: which two channels, which audience first, what to exclude. A plan generated without those inputs will read well and commit to nothing, which is the same failure as the human version.
What's the most common mistake?
Too many audiences and too many channels. Both come from reluctance to exclude, and both guarantee that nothing gets enough resource to work.
How do I know if the plan is working?
The assumptions list. Compare actual results against what you wrote down. If an assumption was wrong, that's information. If you didn't write any down, results are just noise.