Can you still sell online courses and information products now that AI answers any how-to question free, in seconds? Yes, and the earnings say so, provided you understand what customers are actually paying for, because it changed. Tim Ferriss publicly reported on tim.blog (June 12, 2026) that his book catalog is selling roughly 80% below its 2022 level, with the self-help category down 26.3% year over year; across those same years, the figure on our side kept compounding: more than $11.7 billion earned by experts on Kajabi since 2010. When the how-to is free, experts sell the four things AI can't fake: accountability, ownership of outcomes, vouching, and judgment.
That last sentence is the whole argument. The rest of this piece is the evidence, and it starts with the most uncomfortable sales chart in publishing.
Is selling information products dead?
The raw version of it is dying in public, and we can point at the numbers because the man they belong to published them himself.
In June 2026, Tim Ferriss put his own sales data on his blog (tim.blog, June 12, 2026), which is something successful authors do roughly never. His catalog is down about 80% from where it sat in 2022. The self-help category as a whole dropped 26.3% year over year. These figures are self-reported, so apply whatever discount you like... though authors rarely invent numbers that make themselves look worse.
His diagnosis matched what we'd been watching from our side of the fence. For years, buying his book was the most efficient way to get his advice. Now the same question goes into a chat window, and the answer comes back fitted to the asker's own week, their own budget, their own bad shoulder... assembled from his catalog and everyone else's. Before the kettle boils. Free.
And look at what that did to distribution. Google captured demand but passed some of it along as clicks. An AI assistant answers the whole question inside the chat window, and nobody clicks through to anyone.
Ferriss writes how-to at a level almost nobody alive can match, and the how-to is what stopped selling. If raw information were going to hold its price anywhere, it would have held it there.
Then why did expert earnings keep compounding?
Because the money on our platform has always gone to people selling something other than raw information, even when their offers looked like information from the outside.
Here's our side of the ledger: more than $11.7 billion earned by experts on Kajabi since 2010... their earnings, to be clear, not our revenue... across hundreds of thousands of experts. That number kept compounding straight through the years Ferriss's chart fell off a cliff.
Two possible readings. Either expert businesses are the exception to the how-to collapse, or they're simply later on the same curve. We've argued both sides at our own whiteboard, so here's the honest version: the shape of the earnings leans hard toward the first reading, because of who the money goes to.
Among experts who have made at least one sale, the money has always been lopsided in a specific way. Line every one of those sellers up, poorest to richest. The seller halfway down the line has earned $3,653. Total, lifetime. The average across that same line is $122,047, a number that describes almost no one, because the top end earns enough to drag the average into the stratosphere... the top 1% hold 44.3% of the total, and about 14% cross $100,000.
We can't hand you a controlled experiment explaining why (nobody can), but we can tell you what we see when we look at who sits on the rich end of that line, year after year. Their offers have a human obligation attached. A cohort with a start date and someone who follows up when you go quiet. A program that promises a specific result and stands behind it. A certification with a bar you have to clear. The people selling shrink-wrapped explanation cluster near the middle of the line, and they clustered there before ChatGPT existed. Their model was already weak. AI removed the last reason anyone had to pay for it.
The "creator economy" everyone named a decade ago was always the visible sliver of a bigger thing: the expert economy, where the product is what someone knows and what they'll stand behind. The four scarce things below are what that economy actually sells.
What do experts sell now?
Four things stayed scarce when information went free. Each is something information never was, each is something an AI can't credibly offer, and each is a business model we watch working at scale. We wrote a full piece on each one; here's the shape of all four.
Accountability
Ask an AI for a marathon training plan and you'll have a great one before your coffee cools. Running it for sixteen weeks is a different product. The plan went free, so the paid thing became a human who expects you at Thursday's check-in, adjusts the plan when your knee complains, and follows up when you disappear in week nine. Coaches felt this shift first because their market forced them to. We covered why accountability became the thing coaches actually charge for in its own piece.
Ownership of outcomes
Advice is now free in every direction you look. Responsibility still costs real money. When a consultant says "I'll get your team through this migration, and my name is on the result," the client is buying risk transfer... someone besides themselves on the hook. An AI will generate ten recommendations in a minute and feel nothing when all ten fail. Here's how consultants sell ownership of the outcome instead of advice.
Vouching
An AI can teach anyone anything. It can't stake a reputation on them. When an expert certifies a practitioner... trained under me, meets my bar, my name travels with hers... the student is buying the right to borrow credibility that took a career to build. Free tutorials made that more valuable, because now anyone can claim they learned the skill somewhere. Here's why vouching is the real product behind certification.
Judgment
When content was scarce, producing more of it paid. Content is now infinite, so filtering it is what pays. A creator with actual taste, telling a specific audience which three things matter this week and which forty to skip, is selling judgment: a point of view with a track record you can check, attached to a person with something to lose by being wrong. An AI will summarize anything you hand it and has no stake in any of it. Here's how creators get paid for judgment when content is infinite.
Who dies in this molt, and who thrives
Here's the stand, and we know it sounds harsh: AI killing the price of raw information is the best thing that's happened to actual experts in a decade.
The experts getting hurt are the ones whose entire product is explanation. The forty-video course that explains things. The PDF vault. The membership that's a content archive with a login screen. Their moat was the effort it took to gather and organize the information, and that effort now costs a prompt. A lot of that market, frankly, was arbitrage... people repackaging what Google already knew into something that looked proprietary. AI deleted the arbitrage. If you spent twenty years earning your expertise the slow way, your cheapest competitors just left the market. (You're welcome to feel bad for them. We don't, much.)
I'll speak for myself here. The first thing I ever sold online was a set of instructions for building a backyard water toy out of PVC pipe. Raw information, and it's the reason Kajabi exists. Today an AI writes you that build list in ten seconds, with a shopping run mapped to your nearest hardware store. I would have needed a better offer. Every expert now starts where I would have started, and honestly, the better offer was always the better business.
"But my course IS information. Am I finished?"
Only if you leave it built that way, and only if you keep treating the course as the whole business. A course is packaging... one of five models an expert business can run on, next to coaching, community, consulting, and certification, and choosing among the five business models for selling your expertise is its own decision. Every one of the four scarce things can be wired into an offer you already have. Give it a start date, a cohort, and check-ins: now it carries accountability. Promise a defined outcome and stand behind it: ownership. Assess people at the end and put your name on the ones who pass: vouching. Cut it down ruthlessly to what one specific person needs and skip everything else: judgment. The information stays in the offer as the cheapest ingredient, with the scarce thing built around it.
Where to start
We went further on each of the four in its own piece, with the data we have and the caveats it deserves. Start with accountability, the clearest case of a free plan and an expensive follow-through, then work through ownership, vouching, and judgment. By the fourth one, you'll know which of them your expertise should be packaged around... and that decision matters more than any tool, platform, or format you'll pick afterward.
