Marketing with AI, for small teams.

AI for marketing pays off fastest in the work between an idea and a published post: sorting customer feedback, planning a month of content, turning one piece into ten, and writing ad variations to test. It is weakest where marketing is won: knowing what customers actually care about, and having proof. Keep people on strategy, facts and anything that claims to come from a customer.

This playbook, written for teams of one to five, covers six jobs to hand off with a prompt for each, the rules on endorsements and AI labels, and a weekly routine.

01AI for marketing: what to hand off, what to keep

Small teams do not lack ideas. They lack hours. AI gives hours back on the production work, while the judgment stays with you.

02Start with customer research, not content

A lot of weak AI marketing starts with “write me a post about our product.” The model knows nothing about your customers, so it writes for everyone and persuades no one.

Feed it real words from real customers instead: reviews, support emails, sales call notes and survey answers. Remove names and contact details first. Then ask for patterns:

Below are [number] customer reviews, support messages and survey answers about [product or service].
Group them into themes: the problems people had before buying, the outcome they wanted, their doubts before buying, and what they praise most.
For each theme, quote three exact phrases customers used. Do not paraphrase the quotes.
Then list the five phrases that appear most often, in the customers' own words.
Feedback:
[paste, with names and emails removed]

Those exact phrases become your headlines and hooks. They sound like your customers because they are your customers.

Next, use the themes to test your positioning, the one-line answer to “why you, and not the alternative”:

Using the themes below, write three positioning statements in this form: For [who] who [need], [product] is the [category] that [main benefit]. Unlike [main alternative], it [key difference].
Then attack each one as a skeptical buyer would: what is vague, what is unproven, what a competitor could also claim.
Themes: [paste the output above]

Pick one yourself. Then say it out loud to three customers and watch their faces.

03Write one brand brief and reuse it everywhere

A brand brief is a one-page source of truth that you paste into every marketing prompt. It keeps drafts on voice and, just as important, keeps them from inventing claims.

The “claims we can prove” line is the guardrail that matters most. Tell the model to use only those claims, and to write [CHECK] wherever it wants a fact you did not give it.

04Plan a month of content in one sitting

With the themes and the brief, a content calendar takes one conversation. Build it around one strong piece a month, such as a guide, a customer story or a short video. Everything else comes from it.

[Paste the brand brief]
Plan next month's content for [channels, e.g. blog, email newsletter, LinkedIn, Instagram].
Build it around one main piece on [topic from the customer themes]. Everything else should point back to it.
Output a table with: date, channel, topic, hook (first line), call to action, and the asset needed (photo, chart, short video).
Mix formats: how-to, customer question answered, behind the scenes, and one offer per week at most.
Use only claims from the brief. Write [CHECK] where you need a fact I did not give you.

05Create once, repurpose everywhere

Repurposing is where AI saves the most time. One good article or recorded talk holds enough material for weeks of posts, if someone does the slicing.

[Paste the brand brief]
Here is our main piece for the month: [paste the text or transcript].
Turn it into:
1. Five social posts for [platform], each built on one idea from the piece, under [number] words.
2. One newsletter section under 150 words, ending with a link to the full piece.
3. Three scripts for 30-second videos: hook, one point, call to action.
Keep every fact exactly as it appears in the piece. Do not add new statistics.

Then read each output as a customer scrolling past. Cut anything that could have been written by any company in your industry. For the newsletter, our guide to writing emails with AI shows how to keep drafts in your own voice.

06Ads and search content: test, do not guess

AI is very good at writing many versions of an ad. Your job is to make sure the versions differ in a way worth testing, not just in word order.

[Paste the brand brief]
Write ad headlines for [product], [number] per angle, each under [character limit] characters:
1. The problem, in the customer's words
2. The outcome they want
3. Proof (use only claims from the brief)
4. The main objection, answered
Then pick the strongest two from different angles and explain why in one line each.

Run the two winners against each other and change one thing at a time. The same goes for the page the ad points to. Our guide to a landing page that converts walks through each section with a prompt.

Search content needs the same discipline. AI answers now sit above many search results, so a page has to offer something worth citing, such as your own data or experience. Our guide to SEO in the age of AI covers what the studies show.

07The guardrails: endorsements, reviews and AI labels

AI makes it easy to produce content that breaks rules you may not know exist. Three sets of rules matter most for a small team. This is a summary, not legal advice.

Endorsements must be disclosed. In the US, the FTC says that when someone promoting you has a connection your audience would not expect, such as payment, free products or a job, it should be disclosed clearly. Starting a post with “Ad:” or “#ad” would likely work. A disclosure at the end of a post is easy to miss, and a simple “thanks” to the brand may not tell people the product was free.

Built-in platform labels help, but the FTC puts the responsibility on the brand and the creator, not the platform. If you work with influencers or affiliates, it expects you to train and monitor them. Employees who post about your products should disclose the relationship too, and listing the employer on a profile page is not enough.

Fake reviews are banned, AI-written ones included. The FTC’s rule on reviews and testimonials, in effect since October 21, 2024, bans reviews that claim to come from someone who does not exist, such as AI-generated fakes. It also bans undisclosed reviews by staff, paying for reviews on condition that they are positive or negative, and buying fake followers or views. Courts can impose civil penalties for knowing violations.

The FTC notes the rule has no blanket ban on AI-generated avatars in marketing. Presenting one as a real customer is another matter, because the rule bans testimonials from someone who does not exist.

Realistic AI media needs a label. In the EU, the AI Act’s transparency rules have applied since August 2, 2026. If you publish a deepfake, meaning realistic AI-made or altered images, audio or video, you must disclose that it is artificial. For clearly artistic, satirical or fictional work, the disclosure can be light enough not to spoil it. Our guide to the EU AI Act for small builders explains who is covered.

Platforms set their own rules. YouTube requires a label for realistic AI content, and names a promotional travel video with AI-generated footage of a real place as an example. Google’s Merchant Center requires online stores to tag AI-generated product images in their metadata, and to label AI-written product titles and descriptions.

08A weekly AI marketing routine

Tools help less than habits. This loop keeps a small team shipping without letting AI run unchecked.

1. 01

   Review last week’s numbers

   Export last week’s traffic, sign-ups, email results and ad spend into a spreadsheet. Ask AI what changed and why, with each cause marked as a guess to check. Ask it to show its math, and spot-check one figure yourself.

2. 02

   Pick this week’s focus

   Choose one theme from your customer research and one goal, such as sign-ups for a webinar. Everything this week serves that goal.

3. 03

   Draft from the brief

   Generate the week’s posts, email and ad variations from the brand brief and the main piece. Keep a [CHECK] list of every fact to confirm.

4. 04

   Edit, check and label

   A person reads everything before it goes out. Confirm each fact, cut generic lines, add disclosures for any paid or gifted post, and label realistic AI media.

5. 05

   Test one thing

   Run one test: two headlines, two subject lines or two images. Write down the result in the brief, so next month’s drafts start smarter.

09FAQ

Next, learn how to get your site cited by AI search, or make better product shots with AI product photography.