How to Learn a New Marketing Skill With AI in 30 Days

You learn a new marketing skill with AI by making the assistant your tutor, not your ghostwriter. It builds the syllabus, explains the parts you don't get, and marks your homework. You do the work, on a real task, every week. After 30 days you have a skill you can use without the assistant open.

Most people use AI the other way round. They ask for the finished thing, paste it in, and learn nothing. That works until the output is wrong and you can't tell. The point of learning the skill is that you can tell.

This article gives you the method, a four-week plan and the prompts. It uses SEO as the running example, but the same structure works for copywriting, email, analytics or paid ads.

Step 1: pick the skill from the gap in your T

If you have read our piece on the T-shaped marketer, you already know your shape: one deep skill and a bar of working knowledge across the rest. The skill to learn next is not the one that sounds most interesting. It is the weakest point in that bar that costs you something this month.

Three questions find it fast:

1. Which marketing job do you avoid, even though you know it matters?
2. Which job do you pay someone else for, without being able to judge their work?
3. Which job did you do last month that you know was below standard?

The answer that shows up twice is your skill. For a lot of solo marketers it is SEO (the site gets no search traffic), copywriting (the landing page does not convert), email (the list exists but nobody writes to it) or analytics (the numbers are there but nobody reads them).

Then narrow it. "SEO" is too big for 30 days. "Write and optimize one article that ranks for a question my customers ask" is the right size. You are learning the slice of the skill you need for one real outcome.

Step 2: ask the AI for a syllabus, not a lecture

Open Claude, ChatGPT or whichever assistant you use. The free plans are enough; Claude Pro and ChatGPT Plus are $20 a month each if you want longer sessions (checked September 2026). Start a new chat, or a project if your assistant has them, and keep the whole 30 days in it so the context builds up.

Paste this:

I'm a [your role] who does my own marketing. I want to learn [skill] well enough to [the real outcome] in 30 days, at about 30 minutes a day on working days. What I already know: [two or three honest lines]. What I don't know: [two or three lines].

Build me a four-week syllabus. Each week must end with one piece of real work I ship, not an exercise. For every week, list: the three concepts I need, the primary source where each one is documented (official docs, not blogs), the task, and how I'll know I did it well. Don't teach anything yet. Ask me three questions first if my goal is unclear.

The last line matters. A good tutor checks what you mean before teaching, and asking for questions first gets you a syllabus fitted to you instead of a generic course outline.

Read what comes back and cut it down. If week one has seven concepts, ask for three. If a task is an exercise ("write five practice headlines"), ask for a real one ("write the headline for my actual page").

Step 3: learn by doing a real task each week

The syllabus is only useful if every week ends in something you ship. That is what makes this different from a course: the practice is your marketing.

Each day, split your 30 minutes in two. Spend ten minutes learning one concept, and twenty minutes applying it to the week's task. Use the assistant as a tutor during the ten minutes with this prompt:

Teach me [concept] as it applies to [my task]. Explain it in plain words, give me one example from a business like mine, then ask me two questions to check I understood. Don't move on until I answer. Name the primary source for any rule you state.

ChatGPT has a built-in study mode that works this way, asking guiding questions instead of giving the answer; OpenAI lists it for Free, Plus, Pro and Team plans (checked September 2026). The prompt above does the same job in any assistant.

The rule for the twenty minutes: you write first. Don't ask the assistant to draft the thing you are learning to make. If you are learning copy, you write the headline. If you are learning SEO, you pick the keyword and write the title. The assistant comes in after, as the reviewer.

If you want the other mode, where the assistant does the work and you review it, that is a different skill with its own method. Our guide to delegating your first marketing task to AI covers it. Delegation is for tasks you already understand. Learning comes first.

Step 4: get feedback on your own work

This is where an assistant is most useful as a tutor. A course cannot look at your draft. The assistant can, at any hour, as often as you like.

The trick is to give it a standard to mark against, or it will praise everything. Ask for the standard first, then the feedback:

Here is my [draft / plan / setup] for [task]. Before you comment, write the five criteria a senior [SEO specialist / copywriter / email marketer] would use to judge it, with the source for each. Then score my work 1 to 5 on each criterion, quote the exact line that cost me points, and tell me the one change that would improve it most. Don't rewrite it for me.

"Don't rewrite it for me" keeps you in the driver's seat. You fix the draft yourself, paste it back and ask for a second score. Two or three rounds of that on real work teach more than a week of reading.

Keep a running file of the corrections you get more than once. By week three it is your personal checklist for the skill, and the start of a prompt you will reuse. The ten prompts solo marketers actually reuse all started as checklists like this.

Step 5: check the AI's claims against primary sources

An AI tutor explains well and is sometimes wrong with total confidence. For marketing skills that often shows up as outdated platform rules: a character limit that changed, an SEO tactic Google now discourages, an ad setting that was renamed.

So every rule the assistant teaches goes through one check: where is this written down by the people who control it? For each skill, that means:

• SEO: Google Search Central documentation and the Search Console help center.
• Paid ads: the ad platform's own help center (Google Ads, Meta, LinkedIn).
• Email: your email provider's docs, plus Google's and Yahoo's sender guidelines.
• Analytics: the Google Analytics help center, or your tool's docs.
• Copywriting: no official source, so the check is your own numbers. Test the claim on real traffic before you believe it.

A useful prompt at the end of each week:

List every factual claim you made this week about how [platform] works. For each one, give the source and say how confident you are. Mark anything you can't point to a primary source for.

Then open the three claims you depend on most and read the source yourself. Once or twice a month you will catch something wrong. That catch is part of the skill: you now know enough to notice.

The 30-day plan (worked example: SEO for a small site)

Here is what the plan looks like for someone learning to get one article ranking. Swap in your own skill; keep the shape.

Week 1: understand the ground

• Concepts: how Google finds and ranks a page, search intent, what Search Console shows you.
• Primary sources: Google's SEO Starter Guide and the Search Console help center.
• Task you ship: connect Search Console if you haven't, and write down the ten queries you already get impressions for.
• You did it well if: you can explain, in your own words, why your top page ranks where it does.

Week 2: choose the target

• Concepts: keyword intent, the gap between what you rank for and what customers ask, reading a search results page.
• Task you ship: one target question, with a one-paragraph brief: who searches it, what they want to find, what the current top results miss.
• Feedback round: paste the brief and ask the tutor to score it against the criteria prompt above.

Week 3: write and optimize

• Concepts: title and description, headings that answer the question, internal links.
• Task you ship: the article itself, written by you, reviewed by the assistant, fixed by you.
• Source check: Google's guidance on title links and snippets, before you publish.

Week 4: measure and teach it back

• Concepts: what to watch in Search Console in the first weeks, what not to panic about.
• Task you ship: a one-page checklist of how you do this skill now, written in your words.
• Final test: ask the assistant to quiz you with ten questions, no hints. If you get eight, you have the working level.

The teach-back in week four is not optional. Writing the checklist forces you to find the parts you only half understand, and it is the thing you will open next time.

If you would rather start from a ready-made four-week plan for your goal, the Plan Builder gives you one for common goals like ranking a small site or starting a newsletter, with a tool per step. Pair its steps with the tutor method here and each week becomes a lesson too.

The 15-minute version: start today

You don't need a free month to begin. Do this today:

1. Answer the three gap questions from step 1 and write down one skill and one real outcome.
2. Open your assistant and paste the syllabus prompt from step 2.
3. Answer its three questions honestly.
4. Cut the syllabus to three concepts a week and one real task per week.
5. Put a recurring 30-minute block in your calendar for the next four weeks, named "[skill]: tutor + task".

That calendar line is the part people skip. The assistant is patient forever; your good intentions are not. A standing block with the skill in its name is what gets you to week four.

Where this fits in a one-person marketing team

Adding one skill a month is how a solo marketer covers more of the eight jobs a team would normally share. Our book The One-Person Marketing Team maps those jobs and shows how to turn each into a system you run in a few hours a week. It is a good place to see which gap matters most before you pick the skill.

And if you are new to AI assistants altogether, learn the assistant before you learn a skill with it. Your AI Colleague is a five-day email sprint, fifteen minutes a day, that covers briefing, checking facts against the source and building the habit. It works on Claude's free plan, and the fact-checking day is the same muscle step 5 depends on.

Common mistakes

Learning two skills at once. You will finish neither. One skill, 30 days, then decide on the next.

Letting the tutor write the homework. If the assistant drafts the thing you are learning, you have learned to prompt, not the skill. Write first, always.

Asking "is this good?" You will get a yes. Ask for criteria first, then a score, then the one biggest change.

Skipping the source check. The assistant's confidence is not evidence. The platform's own documentation is.

Practicing on fake work. Exercises are forgettable. A real page, a real email or a real report gives you results to learn from, and it does your marketing at the same time.

Thirty days from now you can have one more real skill in your T, a checklist in your own words, and a piece of marketing that exists because you learned how to make it. Start with the gap.