Back to catalog
AI at workBeginner

AI for Marketing

Copy, campaigns, audience and analysis with a language model — and the places where it quietly costs you customers

AI for Marketing

About this course

A course for the person who has to ship the copy: a marketer in a small team, a founder writing their own landing page, a copywriter who now works twice as fast and is not sure that is a good thing. It starts with the most common disappointment — you ask for a selling text, you get four fluent paragraphs that could belong to any company in your industry — and takes apart why that happens and what to hand the model instead.

Along the way you assemble the materials that make every later request better: a product description, a written description of your brand voice, a sheet of facts and figures you can actually prove, and a list of things that must never appear in your copy. Then the channels: a landing section, an email, a social post and an ad are four different jobs, and the differences are specific enough to put into a prompt. Then the inputs — reviews, feedback and competitors' public pages — and the volume questions: a content plan, repackaging one piece of work into several, and why mass generation for search is a bad bet.

The last module is the uncomfortable one. Fabricated reviews, invented case studies and numbers added for effect are not clever marketing; they are a lie to the buyer and a conversation with a lawyer waiting to happen. And a text that sounds good is not a result — this course ends with how to tell whether anything you shipped actually worked. No programming, no dependence on any particular model or advertising platform.

Course curriculum

  1. The brief, not the wish

    Why a request to write a selling text comes back empty, what the model actually needs to know about your product, and which materials are worth assembling once and reusing in every request afterwards.

    • 1.Why 'write a selling text' comes back empty12 min
    • 2.The materials you assemble once14 min
    • 3.Where generated copy quietly costs you12 min

  2. Copy that has to work in a channel

    A landing section, an ad, an email and a social post are four different jobs with four different readers. This module turns those differences into instructions, and ends with variants worth testing.

    • 1.Landing pages and ads: short forms under constraint15 min
    • 2.Email and social: a sequence and a feed14 min
    • 3.Variants for A/B testing, without ten synonyms12 min

  3. Listening: customers and competitors

    Two jobs where a model is genuinely strong — sorting hundreds of reviews into categories and harvesting the words customers actually use — plus an honest account of what competitors' public materials can and cannot tell you.

    • 1.Reviews and feedback: categories and the customer's own words14 min
    • 2.Competitors, read from their public materials12 min

  4. Volume, languages and honesty

    Producing a lot of content without producing landfill, moving marketing copy into another language and culture, and the line between persuasion and deception — plus how to tell whether any of it worked.

    • 1.Content at volume: the plan, the repackaging, and the SEO trap15 min
    • 2.Localisation: translation is the easy half14 min
    • 3.Honesty and results: what is a risk, and what is a metric16 min