The argument in one line.
The course business model is permanently dead due to free YouTube content and AI that can do the work instead of teaching people to do it, so successful information publishers must pivot to selling AI tools trained on their IP, offering services, or building convention events.
Read if. Skip if.
- You built a 6-7 figure info product business (courses, certifications, memberships) and are seeing sales decline or stagnation over the past 12-24 months.
- A content creator or educator with an established audience and intellectual property who wants to understand what replaces the course model in a saturated market.
- You run a digital marketing agency or education company and are considering a pivot away from course sales but don't know what the replacement revenue streams are.
- You're still in growth mode with your course business and haven't yet experienced the saturation and sales decline this breakdown addresses.
- You're interested in how to BUILD a course from scratch or optimize an existing course that's performing well — this is about why the model itself is broken, not how to execute it.
- You operate in a niche market (certification, compliance, specialized B2B training) where courses remain a viable business model and haven't faced the commoditization Deiss describes.
The full version, fast.
The course business model is structurally broken: YouTube saturated the supply of free instruction and AI convinced buyers they can skip learning entirely and just have the work done, which is why one eight-figure education company saw sales collapse to 20% of prior levels. The replacement play treats your business as a publishing company with four monetization paths � premium content, advertising, events, and licensing � and pivots course IP into AI assistants, GPTs, prompt packs, or productized services that deliver the outcome instead of teaching it. To scale toward eight figures, build owned media first, then layer in a convention-style event that becomes the place an industry gathers, expecting three to four unprofitable years before sponsors chase you and ticket-plus-sponsorship revenue splits roughly fifty-fifty.
Chat with this breakdown.
Modern Creator members can chat with any breakdown — ask for the hook, quote a framework, find the exact transcript moment. Unlocks at T2: refer 3 friends + add your own API key.
Create a free account →Who's talking.
Where the time goes.

01 · Cold open montage
Teaser clips of Ryan's most alarming quotes — drops to 20%, took it down to the studs.

02 · Why Ryan retired from DigitalMarketer
25-year history of digital publishing, the April 2025 decision to stop investing in new courses, the 'slow and profitable death' framing.

03 · The real reason course sales collapsed
Three headwinds: post-COVID bubble pop (Clubhouse analogy), YouTube as free-education substitute, AI as the killing blow.

04 · Why AI beats courses every time
AI doesn't have to be better — just believable enough. People take path of least resistance. If AI can plausibly do the thing, the course is dead.

05 · The 80% revenue drop
The DM numbers land: dropped to 20% of prior revenue in 12-18 months. Matt contextualizes DM's scale.

06 · Course business is the wrong frame — it's publishing
Reframing as publishing unlocks century-old business models. Publishers vs authors vs distributors. Category determines who you benchmark against.

07 · Build your own media first
All monetization paths require owned media. Social to email. Being seen vs being followed. Unique POV driven by unique experience.

08 · The 4 ways media companies monetize
Advertising, premium content, events, licensing. Most people stuck on #2 (courses).

09 · Sell AI trained on your IP
Play 1: GPTs/skills/prompt packs built on your course content. DM's actual pivot. Play 2: Productized services — use the AI to deliver the end result for clients. Scalable.co's 12-16 week OS-build engagement.

10 · Why coaching on top of courses won't save you
New audiences want the end result, not learning. Productized services work because they sell outcomes. Course+calls at 10x only converts existing believers.

11 · The truth about cohort-based programs
Cohorts are a good addition but not a business model. Feast/famine cash flow, scheduling friction, same fundamental problem as courses.

12 · The 3 event business models
Expo (trade show), Convention (content+networking, TNC model), Community/sales event (Hormozi workshop model). Each monetizes differently.

13 · Sponsorships, exhibitors and critical mass
~3,000 attendees before sponsors take you seriously. 50/50 ticket/sponsor split is healthy. Chicken-and-egg years 1-2. Year 3 habit forms, year 4-5 flywheel starts.

14 · Sponsorship tiers and protecting the stage
Work top-down from presenting sponsor. Approve topic AND speaker for any stage time. Better: build the sponsor's presentation yourself. Coffee sponsors, room naming, booth location as tier levers.

15 · Event programming, activations and wrap-up
45-min sessions with consistent start/stop across stages. Breaks critical for exhibitors. Year 1-2 overspend on what your audience values. Where to find Ryan.
Lines worth screenshotting.
- Course sales at a leading digital marketing company dropped to 20% of prior levels — not down by 20%, but to 20%.
- If a consumer believes AI will do what your course teaches them, they will take the AI path every single time.
- AI doesn't need to be better than your course — it just needs to seem good enough to try first, and most people will never come back.
- Stop selling the content you created; train an AI on that content and sell access to the AI instead.
- The publishing industry is a more useful frame for course businesses than 'info products' — and it unlocks completely different business models.
- Courses work until people believe AI will do the thing the course teaches; any skill that reduces to 'generate bits' is now at risk.
- An 8-figure convention event business takes 5 or 6 years to become profitable — the first 3 years, you should expect to lose money or break even.
- Events become the place when people attend because other people are there, not because you marketed it well — that shift takes about 4 years.
- A healthy convention event generates roughly 50% of revenue from ticket sales and 50% from sponsorships.
- Cohort-based courses sit between pure courses and coaching and inherit the scheduling friction of both without solving the core problem: people want the result, not the process.
- Productized services — where the process is identical for every client even if the output is customized — are the most scalable version of a coaching pivot.
- Thirty-minute breaks at events are not wasted time: they are the networking the attendees actually came for.
- The presenting sponsor package should be built first because it defines every deliverable that lower-tier sponsors can subset from.
The course is dead. The publisher is not.
Ryan Deiss just handed you the exact playbook Joe has been building toward: own the media, train the AI on your IP, sell the AI or sell the done-for-you result — never sell the course again.
- Apply the AI diagnostic to every product in your line: could someone reasonably believe ChatGPT will just do this? If yes, pivot now before the market does it for you.
- The 'publishing' reframe is the right identity for MCN+ — not a course platform, a publishing company with AI-enabled products and a services arm.
- The AI-on-IP play is literally what MCN+ should be: take the IP you've built (Direct Response 101, The $6 Stack, etc.), train Claude skills and GPTs on it, sell access to those instead of (or on top of) the course.
- The productized services path is the LFB Line — done-with-you is already in the offer architecture; now AI-enable the delivery so one person can run 10 clients.
- Convention timeline reality-check: if you're building an in-person event, expect 3+ years before sponsors come to you. Invest in attendee experience first, monetization second.
- Use this episode's quotables as content: 'dropped to 20%', 'atoms and not bits', 'AI doesn't have to be better' — three viral hooks already clipped and ready.
Terms worth knowing.
- Info product
- A digital product whose value is the information inside it — courses, ebooks, certifications, training programs — sold as a packaged asset rather than a service.
- GLP-1
- A class of prescription drugs (like Ozempic and Wegovy) originally for diabetes and weight loss that suppress appetite and reportedly dampen compulsive habit loops around food, alcohol, and other consumption.
- GPT (custom)
- A pre-configured version of ChatGPT loaded with specific instructions, knowledge files, or tools so it behaves as a specialist for one task or domain. Creators sell or share access to these for repeatable workflows.
- Claude skill
- A reusable instruction-and-context bundle inside the Claude assistant that turns it into a domain expert for a defined task. Functionally similar to a custom GPT but for the Claude platform.
- Prompt pack
- A curated bundle of prewritten AI prompts for a specific job — writing emails, building ads, analyzing offers — sold or given away so a buyer gets results without learning prompt design.
- Lead magnet
- A free resource — checklist, template, mini-tool — offered in exchange for an email address to start a relationship with a prospective customer.
- Middle-of-funnel content
- Material aimed at people who already know the brand and are evaluating it, designed to deepen interest and move them toward a purchase rather than introduce them cold.
- Productized service
- A done-for-you service packaged with a fixed scope, fixed price, and a repeatable delivery process, so it can be sold and fulfilled almost like a product.
- Cohort-based course
- A course delivered to a group on a fixed schedule with live sessions, deadlines, and peer interaction, rather than a self-paced library a student works through alone.
- Pitchfest
- An event or stage session where speakers spend most of their time selling their own products instead of teaching, often hurting the host event's reputation.
- Expo event
- A trade-show-style event built around a large exhibit floor where buyers and sellers meet. Revenue skews heavily toward exhibitor and sponsor fees rather than ticket sales.
- Convention event
- An event built around stage content and networking, where attendees pay to attend for the sessions and the room of peers. Healthy revenue is roughly half tickets, half sponsorships.
- Convex
- A hybrid event format that combines a convention's content and networking with an expo's trade floor, leaning more toward one side or the other depending on the audience.
- Community event
- A single-stage event where a brand gathers its own audience primarily to sell its products or services from the stage, rather than to monetize through tickets or sponsors.
- Presenting sponsor
- The top-tier event sponsor whose name appears alongside the event title and who receives the largest package of visibility, access, and on-site assets.
- Activation
- A branded experience on an event floor — a giveaway car, a coffee bar, a game station, a meetup — designed to draw attendees, create buzz, and give a sponsor something tangible to attach their name to.
- Ideal client profile
- A precise description of the customer a company most wants to land, used to target outreach, qualify leads, and — at events — match sponsors with attendees worth meeting.
- Owned media
- Audience channels a brand controls directly, like an email list or a membership site, as opposed to rented reach on platforms like Instagram or YouTube where the algorithm decides who sees what.
- BYOK
- Bring Your Own Key — a pricing model where the software is sold or included in a membership, but the customer plugs in their own API key for paid AI services and pays those usage costs directly.
Things they pointed at.
Lines you could clip.
“They didn't drop by 20%. They dropped to 20%.”
“AI doesn't have to be better. People just have to believe it's at least as good, and they're gonna do it first because it's already there, it's pretty much free.”
“Could somebody reasonably believe that AI would just do the thing that you're teaching them to do? And if the answer is yes, your course business is in trouble.”
“If what you're teaching people to do still needs to be done with atoms and not bits, then I think you still probably have a decent shot.”
“Stop trying to sell the content. Train AI, sell access to the AI that's been trained on your content.”
“I looked at it and I said, this is just publishing. That's all that it is.”
Where the conversation goes.
Word for word.
The bait, then the rug-pull.
Ryan Deiss built DigitalMarketer into an eight-figure publishing empire that educated a generation of digital marketers — then watched it fall to pieces in eighteen months. Not a dip. Not a correction. A structural collapse to 20 percent of prior revenue, caused by three forces he now names plainly: the post-COVID bubble, YouTube, and AI. This is the post-mortem, and the playbook for what comes next.

































































