The skill behind this guide: Gabi - Affiliate Marketing Specialist AI Skill. Holds your programme terms, your partner pipeline and your disclosure standard, so outreach and briefs stay consistent instead of improvised - $19, one payment, yours permanently.
View the Gabi skill →Affiliate marketing looks passive and runs like a relationship business with a compliance problem attached. An AI chat is genuinely good at three parts of it - the outreach, the terms written in plain language, and the content structure - and dangerous at two: it invents product specifications and prices, and it will happily write you a testimonial from a person who does not exist. Start with the disclosure rules, because that is the part with financial penalties attached. This is operational guidance, not legal advice; the rules that apply to you depend on your market.
Disclosure first, because it is the part with teeth
Most affiliate guides put disclosure at the bottom in one apologetic sentence. It belongs at the top, because since 2024 it is the only part of affiliate marketing where getting it wrong carries a direct financial penalty rather than a strongly worded letter.
In the US, two instruments matter. The FTC Endorsement Guides (16 CFR Part 255), substantially revised in 2023, say that a material connection between an endorser and a seller must be disclosed clearly and conspicuously. An affiliate commission is a material connection. The Guides themselves are interpretive, which is why the second instrument changed the stakes: the FTC's Rule on the Use of Consumer Reviews and Testimonials (16 CFR Part 465), which took effect in October 2024, is a rule rather than guidance, and rule violations can carry civil penalties. It prohibits fake or AI-generated reviews and testimonials, buying positive or negative reviews, undisclosed insider reviews, suppressing negative reviews to create a misleading picture, and buying fake indicators of social media influence.
The phrase that does the work is clear and conspicuous, and it has a specific meaning: the disclosure has to be hard to miss. In practice that means it sits with the claim rather than at the bottom of the page, it is not hidden behind a "more" link, it is not only in a profile bio or a pinned comment, and for video it appears in the video itself - in the audio and on screen, long enough to read - not only in the description. A platform's own "paid partnership" tag is helpful and is not on its own the whole answer. The disclosure should also be in the same language as the content it applies to.
Outside the US the shape is the same and the labels differ. In the UK, the CAP Code and the ASA expect the ad to be identifiable as an ad before engagement, with "Ad" placed up front rather than buried in hashtags, and the CMA gained direct consumer-protection enforcement powers under the Digital Markets, Competition and Consumers Act. In the EU, the Unfair Commercial Practices Directive covers undisclosed commercial intent, and the Digital Services Act requires advertising on platforms to be identifiable as such.
| Where the link lives | What generally counts | The common failure |
|---|---|---|
| Blog post or review article | A plain line above the first affiliate link, visible without scrolling past the recommendation | A disclosure in the footer, or below a table of contents nobody reaches |
| Long-form video | Said out loud and shown on screen, near the start and again near the recommendation | Description box only, or a two-second caption at the end |
| Short-form video | On-screen text that stays up long enough to read, plus the platform tag | Hashtags at the end of a long caption that is collapsed by default |
| Newsletter | A line at the top of the section containing the link | One global note in the sign-up page that subscribers never see again |
| Short social post | "Ad" or the disclosure at the front of the post, before the fold | Buried after three lines and a "see more" |
| Podcast | Spoken, near the recommendation itself | Show notes only |
None of this is legal advice, and the specifics move. The operational point is that disclosure is a design decision made when you build the template, not a thing you remember to bolt on afterwards - and it is the easiest part of the whole programme to standardise across every partner.
Keeps your programme in one place between conversations - commission model, cookie window, brand-bidding and coupon policy, disclosure standard, and where each partner is in the pipeline - so outreach, briefs and follow-ups come out consistent instead of rewritten from memory every time.
View Gabi - Affiliate Marketing Specialist →Two sides of the same table, with opposite incentives
Almost every generic affiliate guide reads as if brand and creator want the same thing. They do not, and knowing exactly where the interests diverge is what makes the terms worth arguing about.
| The clause | What the brand wants | What the partner wants |
|---|---|---|
| Cookie window | Short, so credit goes to demand the programme really created | Long, so a slow consideration cycle still pays |
| Attribution | Last click, which is cheap and easy to defend | Credit for the content that started the search, which last click never gives |
| Brand bidding | Banned, because paying commission on your own brand-name traffic is paying twice | Allowed, because it converts |
| Coupon and extension partners | Excluded or on a lower rate, because they often attach at checkout to a sale already happening | Full rate |
| Returns and clawback | Commission reversed on refunds and chargebacks | Paid on the order, not on the outcome months later |
| Recurring versus one-off | One-off, capped exposure | Recurring, because the work is front-loaded |
A programme stalls when these are agreed vaguely in a rush and then re-litigated per dispute. Writing them down once, in language a creator can read without a lawyer, is the single highest-leverage hour in the whole channel - and it is the exact kind of translation work an AI chat is good at.
Prompt 1 - programme terms a partner can actually read
Recruiting partners who fit
A programme lives on partner quality, not partner count. Fifty signups who never post are worse than five who do, because they create the illusion of a channel. The pitch that gets a reply does one thing: it explains what is in it for them in the first two lines, in terms of their audience rather than your product.
The trap here is that a model, asked to personalise outreach, will invent the personalisation. It will describe a video they did not make, cite a subscriber count it does not know, and compliment a post it hallucinated. That email does more damage than a generic one, because it proves you did not look.
Prompt 2 - outreach that does not invent anything
Content that converts without burning the audience
Affiliate revenue comes from content that is useful first and recommends second - the comparison, the honest review, the "best X for Y" piece. The craft overlaps heavily with ordinary content work, covered in the blog content guide.
The affiliate-specific danger is factual. Ask a model to compare five products and it will produce a clean table of prices, specifications and feature support, and some of those cells will be wrong with total confidence. In ordinary content that is embarrassing. In affiliate content it is a misleading commercial claim about a product you are paid to sell, which is a different category of problem, and it is the fastest way to lose an audience that came to you precisely for the comparison.
The working method is the same one that fixes citations elsewhere: the model supplies the structure and the questions, you supply the facts.
Prompt 3 - comparison content with the facts left to you
When you paste the filled table back, one more instruction earns its keep: write only from the table, and if a sentence needs a fact the table does not contain, mark it rather than supplying it. That single rule is the difference between a comparison you can stand behind and one that quietly invents a warranty length.
The campaign work that sits around the programme - launch sequences, landing copy, partner newsletters, follow-ups that are not outreach - as ready prompts rather than a persona you load. Useful on either side of the table.
View Sales & Marketing Prompt Library →Picking programmes worth your audience, from the creator side
The commission rate is the least important number in the decision, and it is the only one most creators look at. Three questions decide it instead. Does the product genuinely serve the people who trust you - because a bad recommendation costs more than any single payout. Does the traffic you can send actually convert, or are you sending readers into a checkout they will abandon. And can you verify the claims you are about to make on the brand's behalf, or are you relaying marketing copy you have not tested.
A useful discipline: before accepting, write the honest review you would publish if there were no commission. If it is thin, the programme is wrong for you regardless of the rate.
Prompt 4 - the pre-mortem before you take the programme
Where it fails
- Invented product facts. Prices, specifications, feature support, warranty terms, plan limits. It states them confidently and some are wrong. Every cell in a comparison table comes from you or a source you opened.
- Invented programme norms. Ask what a standard commission rate or cookie window is in your category and you get a plausible number with nothing behind it. Treat it as a search term.
- Fake social proof. A model will write you a glowing testimonial from a customer who does not exist if you ask, and asking is now specifically prohibited conduct under the FTC's reviews rule. It is also the single fastest way to destroy a channel built on trust.
- Fake personalisation. In outreach it hallucinates the detail that was supposed to prove you did your homework.
- Jurisdiction blending. It mixes FTC, ASA and EU requirements into one imaginary global standard. Pin the market in the prompt and confirm the current rule yourself.
- Agreeableness. Ask whether your disclosure is sufficient and you will usually be told yes. Ask instead what a regulator would object to.
- Tracking arithmetic. Do not have it compute EPC, conversion rates or payouts. Have it build the sheet and the formulas; run the numbers yourself.
How do you install the Gabi skill?
The download is a ZIP with SKILL.md at the root of the archive, not inside a nested folder - that folder structure is the usual reason an upload fails. In the Claude desktop app, open Customize → Skills, upload the ZIP and toggle it on. Claude loads it whenever your request matches what it covers.
Skills are available on the Pro, Max, Team and Enterprise plans, with code execution enabled. On Pro and Max you turn on Code execution and file creation under Settings → Capabilities yourself; on Team and Enterprise that switch sits with an owner in organisation settings, which is why the menu is sometimes greyed out on a work account. The full walkthrough is in the skill installation guide.
In ChatGPT or Gemini there is no upload step: open SKILL.md, copy the contents, and paste them into custom instructions. You lose automatic triggering and keep the method.
Who is this for?
Brands and shops running or starting a partner programme, creators and publishers choosing what to promote, and the agency or freelance affiliate managers in between. It works in Claude, ChatGPT or any AI chat.
Affiliate work sits next to several adjacent jobs, and the catalogue keeps them separate rather than bundling them into one vague marketing assistant:
- Zoe - Cold Email Copywriter AI Skill ($19) - the recruitment outreach at volume
- Sasha - UGC Brand Pitch Agent ($39) - the creator side, pitching brands rather than being pitched
- Maya - Social Media Content Writer AI Skill ($19) - the short-form content the links live in
- Teo - Video Marketing Specialist AI Skill ($19) - where most disclosure mistakes happen
- Natalie - Public Relations Specialist AI Skill ($19) - for when a partnership becomes an announcement
The wider sets are the sales & marketing growth skills and the AI marketing agents, and the shortlist for this whole area is in the best Claude skills for marketing. If you want the marketing function as running agents rather than one skill, the Marketing Agent Bundle is five agents for $99 instead of $195 bought separately.
In summary:
Design the disclosure into the template before you recruit anybody, write the programme terms down once in language a creator can read, and never let a model supply a product fact or a testimonial. Use it for the outreach, the terms and the structure; supply the facts yourself. For the programme engine, Gabi - Affiliate Marketing Specialist AI Skill ($19); for the campaign work around it, the Sales & Marketing Prompt Library ($9). Both work in Claude, ChatGPT and any AI chat, with a 30-day money-back guarantee. Operational guidance only, not legal advice.
Gabi - Affiliate Marketing Specialist AI Skill
One file that holds your commission model, cookie window, brand-bidding and coupon policy, disclosure standard and partner pipeline, so recruitment, briefs and follow-ups come out consistent instead of improvised. No subscription. Yours permanently.
KissMySkills is a marketplace of 1,900+ AI skills, 60+ prompt packs, 90+ agents & free tools for Claude, ChatGPT & any AI chat.
Related Skill Guides
- The Best Claude Skills for Marketing
- How to Use Claude to Write Blog Content
- How to Use Claude for Email Marketing: The Nadia Skill Guide
- How to Use Claude as a Growth Marketing Strategist: The Sofia Skill Guide
- How to Use Claude for Marketing Analytics: The Rafa Skill Guide
- How to Use Claude as a Predictive Marketing Analyst: The Tara Skill Guide
- How to Install a Claude Skill: The Complete Guide
