A marketing plan is a Skill job: you give the AI your goal and your hours, and in one sitting it drafts a ninety-day plan that runs one channel at a time. You spend about an hour on it, most of that on gathering inputs and cutting the draft down to what you will really do. This page covers the whole marketing plan, every channel that brings in customers, and treats each channel as a turn in the order. If social media is the channel your plan picks, our social media strategy page carries the six decisions for that channel.
The plan you want is short. One goal, one channel at a time, a weekly task for each week, and one number to read at the end of each month. Anything longer is usually a plan nobody opens again after week two.
- Shape
- Skill
- Sitting
- one sitting
The shape: Skill
This job is a Skill because the work is thinking, and the AI is good at structured thinking when you hand it real constraints. There is no tool to connect, no account to set up and nothing to publish. You paste a brief, read what comes back, push back twice, and keep the version that fits your week. Every ninety days you paste the same brief again, with what happened, and write the next plan.
What the AI cannot do on its own is know your business. It does not know which customers pay on time, which season is slow, or that you lose every Thursday to deliveries. That is why the inputs matter more than the prompt. A plan written from a goal and a real number of hours comes back specific. A plan written from "make me a marketing plan" comes back as a list of every channel that exists.
The other half of the shape is the rule that makes the plan work: one channel at a time. A small business rarely stalls at marketing because it picked the wrong channel. It stalls because it started four at once, ran each at a quarter of the effort, and could not tell which one did anything. A plan that starts one channel, gets it to a routine, and only then adds the next is a plan a business with eight spare hours a week can actually run. If you are not sure which channel goes first, this piece on picking one channel covers that choice.
What you give the AI
Gather these before you open the chat. Each one is short, and most you can write from memory.
- One goal, with a number and a date: "twelve new client bookings by the end of January" works, and "grow the business" does not.
- Your hours per week for marketing: the real number, not the hopeful one, so if it is three hours, say three.
- What you sell and to whom: two or three sentences on your main service, your price range, and who usually buys.
- Where your last ten customers came from: referral, Google search, walking past, a past client returning, a partner or social media, in rough counts.
- What you already have: a website, Google reviews and how many, an email list and its size, a Google Business Profile, photos of your work, and any social profiles with whether they are active.
- Any budget for ads or tools: zero is a valid answer, and it keeps the AI from planning paid campaigns you will not fund.
- The dates that matter in the next ninety days: busy season, slow season, holidays you close for, an event, a price change or a new service.
- What you have tried before and dropped: one line each, with why it stopped, so the plan does not repeat it.
Write these into one document or straight into the chat. Plain notes are fine; the AI does not need full sentences.
What comes back
You paste the brief below with your inputs filled in. It works the same in ChatGPT or Claude.
You are helping a small business owner write a 90-day marketing plan.
Use only the facts below.
If something is missing, ask before assuming.
Business: [what you sell, to whom, price range]
Goal: [one goal with a number and a deadline]
Hours per week for marketing: [real number]
Where the last ten customers came from: [rough counts]
What already exists: [website, reviews, email list size, profiles, photos]
Budget for ads or tools: [amount, or zero]
Dates that matter in the next 90 days: [list]
Tried before and dropped: [list, with why]
Rules for the plan:
1. Run one channel at a time, picking the first from where customers already come from and what already exists, and explain the choice in three sentences.
2. Add a second channel only after the first has a weekly routine, and no earlier than week five.
3. Give every week one task that fits inside the stated hours, with the task, how long it takes, and what is finished at the end.
4. Name one number to check at the end of each month, tied to the goal.
5. Do not invent facts about the market, competitors or results, and do not suggest paid ads unless the budget is above zero.
6. Put the plan in a table: week, channel, task, hours, done when.
7. End with a short list of what the owner must decide or check before starting.What you get back is a short opening that names the first channel and why, a twelve or thirteen week table, the one number to check each month, and a list of decisions for you. Expect the first draft to be close but too busy. Two follow-ups usually fix it:
Cut any week that needs more than [your hours] hours, and move that work later or drop it.
Rewrite the weekly tasks so each one starts with a verb and ends with something finished, like "five posts scheduled" or "ten past clients emailed."
Copy the final table into a document or spreadsheet you will actually open on Monday mornings. A plan that lives in an old chat thread is a plan you will not find in week six.
How to judge it
You can reject a bad plan in a minute. Read it with your calendar open and check these.
- It runs more than one new channel in the first month: the most common failure, so send it back with rule one.
- The hours do not add up: if any single week in the table is over your number, the plan will slip that week and every week after.
- It assumes money you did not mention: paid ads, a new website or a hired freelancer in a zero-budget plan means the AI ignored your inputs.
- It contains numbers you never gave it: market sizes, average conversion rates and claims about what posting daily does are filler, often made up, and should be deleted.
- It could be anyone's plan: if you can swap your business name for a competitor's and nothing else needs to change, the plan is too generic to run.
- The tasks are vague: "build brand awareness" is not a task, and "email ten past clients asking for a review" is.
- There is no number: a plan with no monthly check gives you no reason to change course, so it never does.
A good plan feels slightly too small. That is the point. The plan you finish beats the plan that impressed you on day one.
What stays with you
The AI drafts the plan. Running it, and the judgment calls inside it, stay with you.
- The goal: only you know whether twenty new clients is ambitious or easy, and whether the business can handle them.
- The real hours: the AI believes whatever number you type, so a low true number breaks the plan in week one.
- Where customers really come from: the AI works from the counts you give, so if you guessed, check your booking notes or ask your next five customers.
- Facts about your market: anything the plan says about competitors, local demand or seasons is a guess until you check it.
- The weekly work itself: the plan says "ask every client for a review," and you are the one who asks.
- The monthly read: at the end of each month, you look at the one number and decide whether to keep the channel, change the task or move on, which is the reason the plan exists.
- The next ninety days: when this plan ends, paste what happened back into the chat with the same brief and write the next one.
For a deeper look at timing your content to the months when customers decide, see how to plan a quarter of content around customer timing. If the plan adds social media and you are choosing where, how to choose the right social channels covers which profiles matter for a local business.
An editorial example
This example is invented to show the shape of a good draft.
Say you run a one-person dog grooming van with six hours a week for marketing. Your goal is twenty new regular clients by the end of the quarter. Most of your last ten clients came from neighbors who saw the van, and two came from Google. You have a website, forty Google reviews, not one email address on a list, and a Facebook page you stopped posting on last spring.
A good plan from those inputs starts with the Google Business Profile, because that is where searchers already find you and where the reviews live. Weeks one to four: fill in every field on the profile, add twenty photos of finished dogs, ask every client that week for a review, post once a week. Weeks five to eight: keep the profile routine, and add a neighbor referral card that goes to every client on the route. Weeks nine to thirteen: keep both, and restart the Facebook page with one post a week, because by then the photo library exists. The monthly number is new bookings that name how they found you.
A bad plan from the same inputs starts with TikTok, a newsletter, a podcast guest spot and a paid ad campaign in week one.
Plan length: which one to write
| Plan | Good for | Risk |
|---|---|---|
| Thirty days | Testing one channel you have never tried | Too short to see a result from most channels |
| Ninety days | Most small businesses, one channel at a time | Needs a monthly check to stay honest |
| Twelve months | Businesses with fixed seasons and staff | Often stale before the second quarter |
For most owners reading this, ninety days is the right length. Write the next ninety when this one ends.
Of every job on this list, social media is the one that comes back every week, and Boomp is software that does that one for you. See ten posts made from your website, free
