No platform in this guide will take the money for you, and that is a policy choice rather than a missing feature. Meta's commerce policies state that commerce content may not promote the buying, selling or trading of downloadable digital content, digital subscriptions or digital accounts, so a course cannot sit in a Meta Shop or carry a product tag the way a jar of face cream can. TikTok Shop in the United States lists Services and Virtual Products among its prohibited categories and keeps Virtual Goods in an invite-only tier a seller cannot self-apply into, and TikTok Shop is not available in Canada. LinkedIn's published ad format list contains no shop, catalog, product tag or checkout at all. YouTube's product tagging runs through a connected store platform and is limited to creators based in a short list of countries. Those claims were read against each platform's own policy pages in late September 2026, and such pages change without notice, so re-read them before building on any of it. The consequence is the same everywhere: social does discovery and trust, and the transaction happens on your own checkout. Most structural decisions in this category follow from that.

The second thing worth knowing is what the buyer is assessing, because it is not your curriculum. It is whether you can explain something clearly, and whether they will finish. The first is why free teaching content tends to sell better than promotional content, and why one good clip can outwork a good sales page. The second is why the objections that kill course sales are rarely about money. They are about hours per week, whether a human will answer a question at 11pm, and the private embarrassment of buying another thing that sits unopened.

Cohorts and self-paced courses share a word and little else. A cohort has a start date and a seat count, which hands you a real deadline and a real cap. Both give a deliberating buyer a reason to stop deliberating. They also hand you fixed calendar obligations and a fill rate that is either met by a date or lost. Self-paced has neither, so any deadline has to be manufactured honestly, and the binding constraint moves from delivery capacity to sustained attention. Run one playbook across both and one of them quietly underperforms.

Who is buying, and how they decide

Two buyers hide inside this category and they behave nothing alike. The first pays with personal money: the career switcher, the freelancer adding a service, the parent learning a craft, the junior hire chasing a promotion. That buyer follows you for weeks or months before clicking anything, and the purchase is usually triggered by an event rather than by your marketing, such as a layoff, a new tool landing at work, a client project that went badly, or a peer they respect enrolling out loud. They decide on evidence of teaching ability, on whether the format fits their actual week, and often on a single DM reply that proves someone is home. The second buyer pays with an employer's training budget. That purchase is slower and more procedural because it has to survive a manager conversation and a finance process, so it needs an invoice or purchase-order path, a plain statement of outcomes a manager can forward, a receipt with the right entity on it, and sometimes a completion certificate. It tends to be less sensitive to cost and far more sensitive to legitimacy signals: a named instructor with verifiable practitioner history, a real syllabus, fixed dates. Plenty of sellers build only for the first buyer, then wonder why the professional audience they attract on LinkedIn never converts. If both are in your audience you need two checkout paths and two bodies of proof, and you need to know which door each lead came through before you talk to them.

Where this actually sells

Not every platform is worth your time in this category. These are the ones that are, and why.

YouTube People research a skill in words before they pay to learn it, and this is the surface where a buyer will sit with you for an hour before spending anything. That hour is the nearest thing this category has to a free trial, and it is what moves the employer-funded buyer who has to forward something to a manager. A video also keeps enrolling people months later, so the back catalogue behaves like an asset rather than a campaign. Its native selling tools are not the mechanism here. Per YouTube's own help pages in late September 2026, promoting your own store's products requires a connected store platform and the creator to be based in one of twelve listed countries, and the separate YouTube Shopping affiliate program runs in fourteen listed countries that do not include Canada, the UK or the EU. Both lists change, so check them rather than trusting a summary. For a course the mechanism is the description link, the pinned comment and the end screen.
Instagram This is where the private half of the decision happens. Course buyers save teaching posts, sit on them for weeks, then ask one question before committing, and they ask it in the DM rather than in comments. Reels buy reach, carousels carry frameworks people save and return to, Stories carry the enrollment window because they are dated by nature. Three mechanics shape the build. A course cannot be tagged as a product or listed in a Meta Shop, because Meta's commerce policies bar commerce content promoting downloadable digital content or digital subscriptions, so every Instagram sale is a click-out. If you run Instagram lead ads, Meta's developer documentation states the leads generated still belong to the Facebook Page, so you need a Page whether you wanted one or not, and the same documentation gates lead retrieval behind App Review, the leads_retrieval permission and completed Business Verification, and states you cannot retrieve leads while your app is in Development mode. Budget time for that before a launch rather than during one. Messaging runs on a clock too: businesses have up to 24 hours to respond to a user, and promotional messaging beyond that window needs an approved message tag or a paid sponsored message.
LinkedIn For professional and B2B topics the buyer is logged in with their job title attached and their employer's budget within reach, which is why the manager-approval version of this purchase often starts here. The content that works is the document post and the case teardown rather than the hook, because the reader is quietly assessing whether they could defend the spend. LinkedIn has no commerce surface: its published ad format list contains no shop, catalog, product tagging or checkout, so lead capture is the conversion mechanism. Lead Gen Forms support up to three custom questions across a limited field set, which is enough to qualify by role and team size. One hard constraint for anyone promoting a course they do not own: LinkedIn's ad policy states that ads related to affiliate advertising are prohibited, and unlike its restricted categories there is no certification route in, so an affiliate motion here has to be organic.
TikTok Cold reach is fast here relative to the other surfaces, and the trust window is short. Viewers arrive with no prior relationship, so the path to purchase needs more steps rather than fewer, and the job of a clip is to earn a follow or a saved link, not a sale. Two things bind. TikTok's ad policy prohibits get-rich-quick offers in all markets, which rules out most business-opportunity framing in paid creative. And native checkout is closed to this category: TikTok Shop in the US treats Services and Virtual Products as prohibited and Virtual Goods as invite-only, while TikTok Shop is not available in Canada as of late September 2026, with no published launch date. Canadian operators should plan on off-platform checkout and avoid the workarounds traded in creator communities, such as US operator partnerships and bought shop accounts, which put the account, the tax position and the consumer-protection position at risk.
Pinterest Worth real effort for one slice of this category and little outside it: sewing, art, baking, photography, homeschool, planners, home renovation, anything where someone plans a project before they learn the skill. That planning frame suits a course starting in a few weeks better than a feed built for instant reaction, and saved Pins resurface when the project does. The platform also assumes off-platform conversion, which suits a course: Pinterest states that direct links in its ads take users straight to your site rather than to a close-up, that this is the default when creating a new ad and cannot be disabled, and that direct links are supported on Consideration and Leads campaigns across static image ads, standard video, Carousel and Idea ads. That covers a workshop registration funnel without touching catalogs, which Pinterest itself serves a region-unavailable notice on in some markets.

The content that works

The one-problem teaching clip Sixty to a hundred and twenty seconds that fully solves one narrow problem, with no cliffhanger and no instruction to go and find the rest. It works because it is a direct sample of the product: the buyer is trying to establish whether you can explain things, and a withheld ending answers that question badly. These clips also survive being reposted without context, which matters because a first touch often arrives stripped of your profile, your pinned post and your framing.
The full free lesson, long form One real lesson taught end to end, published rather than gated. It does objection work no sales page can: pace, personality, how you handle the messy middle, whether your examples resemble the buyer's situation. It is also the asset most likely to get forwarded to a manager or a spouse, which is where the decision often actually gets approved.
The live workshop or multi-day challenge Treat this as the conversion event rather than the sales page. It creates a dated reason to decide, it lets people experience the teaching format before they buy it, which is the part that matters for cohorts, and it converts on the quality of the teaching rather than the quality of the pitch. Teach honestly, make the offer once and clearly, and attach a deadline that is real.
Student work, not student income Show what people built, shipped, submitted or fixed, in their words and with written permission. For a skills buyer it does more persuading than an earnings claim, and it stays on the right side of the rules. The FTC's Rule on the Use of Consumer Reviews and Testimonials, 16 CFR Part 465, is a trade regulation rule rather than guidance, and it reaches fake or false consumer, celebrity and insider testimonials, undisclosed insider reviews, and incentives conditioned on a review expressing a particular sentiment. Never trade anything for a positive review.
Curriculum in public, including who it is not for Publish the week-by-week map, the prerequisites, the honest weekly time commitment, and an explicit list of people who should not buy. Disqualifying people tends to raise conversion among those who remain and cut refunds, because a large share of refunds in this category are fit problems discovered in week one rather than quality problems discovered at the end.
Practitioner proof Show yourself doing the work you teach: current projects, teardowns, builds, the thing breaking and getting fixed. Buyers in this category are alert to the person whose only visible occupation is selling a course about an occupation. It also keeps generating the raw material you teach from, which is what stops year two of a course business going stale.

The path from scroll to sale

  1. First touch is teaching, not offering. A clip or a searched video fully solves one narrow problem. No link, no pitch, and the only call to action is to follow or save. Judge this layer on reach and saves, never on revenue.
  2. Repeat exposure does the trust work. Profile, pinned content and the next several posts should read as an informal syllabus, so someone who just discovered you can binge the case for you in ten minutes. Most buyers here need weeks, so stop attributing the decision to a single post.
  3. Capture into something you own. Offer a specific artifact, a template, a teardown, a checklist, or a dated free workshop, delivered by DM keyword, bio link or a native lead form. Get the permission properly. Canada's Anti-Spam Legislation prohibits sending a commercial electronic message without consent, and it defines a commercial electronic message as one it would be reasonable to conclude has encouraging participation in a commercial activity as a purpose, whether or not profit is expected. Consent under that Act is not one thing, since it sets out when consent may be implied rather than express, so know which basis you are relying on for each address and keep the record.
  4. Run the live event. A workshop, a multi-day challenge or open office hours where you teach for real and make one clear offer at the end. For higher-commitment cohorts, replace the offer with a short application, because the qualifying conversation is the sale and the application filters out the people you cannot help.
  5. Open a real enrollment window. Start date, seat count, plain deadline, and a page that answers weekly time commitment, refund terms and who should not buy. Checkout on your own site or course platform, because none of these platforms offers a native checkout for a course. Put an invoice or purchase-order path next to the card path so employer-funded buyers are not blocked at the last step.
  6. Onboard within the hour, then harvest. Access, calendar invites, community link and a first assignment small enough to actually finish, because finishing it is what prevents refunds. At the end, collect outcome artifacts with written permission, open the waitlist for the next cohort, and invite your strongest students into a referral or affiliate arrangement, with the disclosure obligation spelled out in writing, because a student promoting for commission is exactly the material connection the disclosure rules reach.

Rules that constrain this category

Policies and regulations change. Treat this as the starting point for your own check, not as legal advice.

  • Courses are largely unrestricted as a product category, with one loud exception: the money-making subcategory is policed hard on several platforms at once. Meta's Community Standard on prohibited commercial practices bars 'Offers of opportunities of unrealistic financial reward for unclear or minimal effort', and Meta's Advertising Standards carry a prohibited commercial practices section pointing at that standard. TikTok's ad policy prohibits get-rich-quick offers in all markets. Pinterest lists 'multilevel marketing models or get-rich-quick opportunities' as an unacceptable business practice. Snapchat prohibits 'Get-rich-quick offers, pyramid schemes, or other deceptive or too-good-to-be true financial offers'. The Unreliable claims section of Google's Misrepresentation policy reaches claims that entice the user with an improbable result, even a possible one, presented as the outcome a user should expect. Teach a skill and you are in open territory. Promise an income and you are advertising in a restricted lane on several platforms simultaneously. Expect automated enforcement to run stricter than the written text on anything income-adjacent, so plan for rejections even where the policy appears to allow the offer.
  • There is no native checkout for this category, and it is a policy choice rather than a gap. Meta's commerce policies state that commerce content may not promote the buying, selling or trading of downloadable digital content, digital subscriptions or digital accounts, so courses, memberships and ebooks cannot be listed in a Meta Shop or tagged as products, although they can be advertised to an off-platform checkout. TikTok Shop in the United States lists Services and Virtual Products among its prohibited categories and places Virtual Goods in the invite-only tier, which a seller cannot self-apply into. LinkedIn publishes no commerce format at all.
  • Geography decides more than tactics, especially for Canadian operators. As of late September 2026 TikTok Shop is not available in Canada and no launch date is published, the YouTube Shopping affiliate program runs in fourteen listed countries that do not include Canada, the UK or the EU, promoting your own store on YouTube covers a shorter list of twelve countries that also excludes Canada, and Instagram's affiliate product tagging in Reels launched in a small set of markets that does not include Canada, the UK or the EU. These country lists change without notice, so verify each on the platform's own help page before building a plan on it, and treat workarounds such as US operator partnerships or purchased shop accounts as account, tax and consumer-protection exposure rather than clever growth.
  • Testimonials are governed by a rule with teeth, not by etiquette. 16 CFR Part 465, the FTC's Rule on the Use of Consumer Reviews and Testimonials, took effect on 21 October 2024 as a trade regulation rule, so the FTC can seek civil penalties against knowing violators. It reaches fake or false consumer, celebrity and insider testimonials, undisclosed insider reviews, review suppression, buying fake indicators of social media influence, and incentives conditioned on a review expressing a particular sentiment, which is the provision that catches 'leave a five-star review and get a bonus' offers. The FTC announced warning letters to ten companies under this rule on 22 December 2025. Separately, its notices of penalty offenses cover Endorsements and Testimonials, Substantiation of Product Claims, Money-Making Opportunities and Education, an unusually exact overlap with this category.
  • Affiliate and paid promotion has to be disclosed in a way a scrolling viewer cannot miss. 16 CFR 255.5 requires disclosure of a material connection the audience would not reasonably expect, and 16 CFR 255.0(f) states that in an interactive medium such as social media the disclosure should be unavoidable, which is what defeats a disclosure buried below a fold or a tap. The FTC's own guidance treats 'affiliate link' on its own as insufficient wording. In Canada the Competition Bureau treats an undisclosed material connection as potentially false or misleading under the Competition Act and expects disclosures to be prominent rather than hidden in long captions, hashtag blocks or a bio, and Ad Standards' current guidance names '#Ad' used on its own as its gold standard disclosure and now asks for disclosure of AI-generated or significantly AI-altered content. Labels are not portable: UK guidance lists '#spon' and '#sponsored' among terms to avoid, so a label the FTC accepts is not automatically safe for a UK audience. On LinkedIn, remember that ads related to affiliate advertising are prohibited outright, so an affiliate motion there has to be organic.
  • Selling into the UK or EU brings cancellation rights that most course-seller advice ignores. Under the UK's Consumer Contracts (Information, Cancellation and Additional Charges) Regulations 2013, the cancellation period for a service or for digital content not on a tangible medium ends fourteen days after the day the contract is entered into, and for digital content the trader must not begin supply before that period ends unless the consumer gave express consent to start and acknowledged that the right to cancel would be lost. Those regulations implement the EU Consumer Rights Directive and comparable rights apply across the EU, but member-state implementation and the fine detail differ, so check the market you actually sell into. Either way, build that consent-and-acknowledgement step into checkout, and publish a refund policy you can honour instead of the blanket no-refunds line that circulates in launch communities.
  • Occupational claims can move you from consumer law into education regulation. Ontario states that career colleges offering vocational programs must, by law, be registered under the Ontario Career Colleges Act, 2005, and that registered operators must have their vocational programs authorised for delivery, with exempt institutions and programs published separately. Other provinces and many US states run their own private-career-training regimes, so check the one covering your buyers. A skills course that makes no occupational claim generally sits outside this. A 'become a certified practitioner' or 'get hired as a' promise deserves legal review before the first ad runs, not after the first complaint.

Selling this without owning it

If you want to sell in this category without making the product, these are the routes that exist. The sourcing directory covers each in more detail.

Course marketplace affiliate programs hosted on Impact (Coursera, Skillshare) Both run their partner programs on Impact's network, so one account covers several catalogues. Coursera's published terms matter more than its headline economics: it commissions purchases of courses, Specializations, professional certificates and Coursera Plus subscriptions, excludes degrees, credits the publisher with the last qualified click where several affiliates referred a sale, and commissions only the first month's transaction on a monthly subscription. Read the live terms rather than a summary, because Coursera also lists a certificate exclusion that sits oddly beside its commissionable professional certificates. Skillshare requires at least one demonstrated channel and audience aligned with its brand, naming blogs, Facebook groups, Pinterest, Instagram, Twitter followings and email newsletters as qualifying channels, plus a Skillshare account and an Impact account. This route suits broad-topic educational content where you cannot vouch for one specific instructor, and it pays on volume rather than on trust.
ClickBank A long-running marketplace for information products, where the refund mechanics are the business model. ClickBank's own policy states it will at its discretion allow the return or replacement of any product within 60 days of purchase, and that sellers may request a refund on a customer's behalf up to 365 days after the purchase date. The clawback is not symmetrical, and this is the part people get wrong: on revenue-share deals the distribution is debited back out of both the seller's and the affiliate's accounts, whereas on cost-per-action deals only the seller is debited and the affiliate is not. Know which structure you are on before you forecast anything, and vet the product yourself, because a marketplace listing is not an endorsement of the claims inside it.
A creator's own affiliate program, run inside their course platform The route with the least distance between your recommendation and the product, because you are promoting one instructor you can actually vouch for. Most course platforms include the tooling: Podia, for example, runs affiliate programs from the same dashboard the creator sells in, supports public or invite-only programs, allows per-product commission settings, and handles tracking and payouts, though it gates the feature to its higher plan tiers, so confirm the creator's plan before you agree terms. Ask for invite-only terms, a co-hosted workshop, a bonus only your audience gets, and a dedicated landing page, since those are the levers that separate a real partnership from a shared link. Two corrections for anyone working from older guides: ShareASale closed on 6 October 2025 with advertisers and publishers moving to Awin, and Shopify Collabs is not accepting new creator signups.
PartnerStack A practical entry point when the education you sell is B2B: certification programs, team training, and the software that surrounds it. PartnerStack runs one partner dashboard across affiliate, referral, reseller and co-sell motions with automated payouts, and its public directory at market.partnerstack.com is where partners find and apply to individual programs. Approval is per program, and our read is to treat B2B partner revenue as a pipeline that pays quarters later rather than weeks later. It fits LinkedIn-native operators whose audience buys with a company card.
Revenue-share delivery partnerships with an expert who has no audience The route nobody lists in affiliate roundups. Plenty of credible practitioners can teach but will not market, which is the inverse of most social operators. Platforms like Maven, still operating and listing expert-led cohorts as of late September 2026, exist because that gap is real, and the gap it leaves you is operational: you bring the audience, the funnel, the enrollment window and the cohort operations, they bring the curriculum and the live teaching. Paper it before the first post: who owns the student list, who carries refund liability, who can sell the recordings afterwards, what happens to the next cohort if either side walks, and whose name goes on the receipt.

What to measure

  • Waitlist additions per week, attributed to the specific piece of content that produced them. A cohort is sold against a date, so a monthly total hides the shortfall until it is too late to fix. This is the earliest honest signal you have.
  • Workshop registration to attendance rate, plus watch depth on the replay. Attendance is a strong leading indicator of enrollment in this category, and a falling show-up rate is usually a reminder-sequence problem rather than a demand problem, which is a much cheaper thing to fix.
  • Days from first touch to purchase, and the first-touch source. Consideration windows here run long, so last-click reporting will credit the DM while the long video did the work. Ask every new student where they first found you and reconcile that answer against your analytics.
  • Lead to enrollment rate split by capture route: DM keyword, native lead form, bio link, video description. These routes convert very differently, and the cheapest leads are frequently the ones that never enroll. Optimising on cost per lead without this split is how ad budgets get wasted in this category.
  • Fill rate against capacity and days to fill, alongside the share of the cohort that came from previous students and their referrals. That share is the health metric for the whole business, because a course business that cannot re-sell to its own graduates is really a paid-traffic business.
  • Refund or withdrawal rate inside your guarantee window, paired with week-one assignment completion. Completion of that first small assignment tends to predict refunds before anyone requests one, which makes it the delivery metric worth watching live rather than reviewing afterwards.

The mistakes that cost the most

  • Building the whole course before selling one seat. Months of production against unvalidated demand, then a launch to an audience that never asked for it. Sell the first cohort from an outline and a date, teach it live, record as you go, and let those recordings become the self-paced product later.
  • Pointing paid traffic at a cohort sales page with no audience and no list. A cohort has to fill by a date, and advertising cannot manufacture the weeks of trust the decision requires. The same budget filling a dated free workshop usually performs better, because the teaching does the converting and you keep the leads who did not buy this time.
  • Marketing with income screenshots. It is a quick route to restricted ad delivery across Meta, TikTok, Pinterest, Snapchat and Google at once, it invites FTC substantiation and testimonial exposure, and it selects for the buyer least likely to finish and most likely to demand a refund. Student work does more work than student earnings.
  • Assuming native commerce exists, then losing a fortnight discovering it does not. No Meta Shop listing, no Instagram product tag, no TikTok Shop listing for a course, no LinkedIn checkout, and for Canadian operators several affiliate surfaces unavailable outright. Build the off-platform checkout, the invoice path and the tax handling first, then start the content.
  • Running the live workshop as a lecture. Fifty minutes of generous teaching, ninety seconds of apologetic offer, no deadline, no seat count, no answer to the hours-per-week question. The event is the sale. It needs one clear offer, a stated cap and a date that closes.
  • Overselling a cohort past what you can deliver. Extra seats degrade the student outcomes the next launch depends on, and in this category the previous cohort tends to feed the next one. Sell out, close the window, open a waitlist, and let scarcity be a fact instead of a tactic.

A course buyer is not paying for information, they are paying for the odds that they will finish, so your content has to prove you can teach rather than prove you got rich.