Side Hustle & Income
Is an Online Course Still Worth Building?
By Jim Vernon, Editor, AI Intelligence International · Published 18 February 2026 · Reviewed against our editorial standards · About the author
Courses are pitched as leverage: build once, sell forever. In practice they are a demanding product with high production costs, aggressive competition and buyers who can now get a decent explanation of almost anything for free.
That does not make them dead. It makes the bar higher and the winning shape narrower than it was.
Key takeaways
- What a course is actually selling: Not information.
- When a course is the wrong format: When the outcome takes under an hour to teach — that is a guide or a template, priced accordingly.
- Validate before recording: Teach the material live to a small paid cohort first.
- Honest numbers: Completion rates for self-paced courses are famously low, often under a fifth.
What a course is actually selling
Not information. Information is free and abundant, and a model will explain any topic patiently on demand.
A course sells sequence, accountability and confidence: someone who knows the path has ordered it, removed the dead ends and will tell you when you are doing it wrong.
If your outline is just an ordered list of facts, you have written a book badly. The structure and the feedback are the product.
When a course is the wrong format
When the outcome takes under an hour to teach — that is a guide or a template, priced accordingly.
When the audience needs the work done rather than the skill learned. They want a service and will resent being taught.
When you have never delivered the outcome for someone else. Teaching from theory is visible to students within two lessons.
Validate before recording
Teach the material live to a small paid cohort first. It validates demand, produces testimonials, and surfaces the questions your recorded version must answer.
Cohort delivery also prices higher than self-paced, which means the validation phase can out-earn the eventual product.
Record only after two cohorts have gone through it and the structure has stopped changing.
Honest numbers
Completion rates for self-paced courses are famously low, often under a fifth. This affects refunds, testimonials and word of mouth, all of which drive future sales.
Refund rates rise sharply with price when the promised outcome is vague. Precise promises reduce both refunds and complaints.
Sales are concentrated around launches unless you have an evergreen traffic source. "Build once, sell forever" usually means "launch repeatedly".
Designing for completion
Short lessons, one action per lesson, and a visible artefact the student produces. People finish things that produce something.
Build in checkpoints — a quiz, a submission, a call — where a student who has drifted is pulled back. Accountability is the feature that free material cannot replicate.
Keep the course shorter than feels right. Length signals value to buyers and destroys completion for students; solve that tension with depth in optional material, not required length.
Where AI fits
Excellent for producing exercises, quizzes, worked examples, alternative explanations and translated captions — the supporting material that usually gets skipped.
Poor at the core teaching, because the value is your specific experience of where students get stuck.
A useful pattern is an assistant trained on your course material to answer student questions between sessions, with your review of anything it gets wrong.
The numbers that decide it
Take an honest estimate: a genuinely good six-hour course takes eighty to a hundred and fifty hours to build the first time, including scripting, recording, editing and the assets people actually use.
At a hundred and fifty for the course, that is a hundred sales to reach a modest hourly rate on the build alone, before promotion, refunds and platform fees. A hundred sales requires an audience or an advertising budget.
Run that arithmetic before recording anything. Most abandoned courses were viable ideas with an unexamined sales assumption.
Validate with a live cohort first
Teach the material once, live, to eight to fifteen people who pay something. You will learn where the explanation fails, which exercises work and what the real questions are, none of which is knowable from an outline.
The recording of that cohort, plus the questions asked, is the script for the produced version. That sequencing turns the most expensive step into the second step rather than the first.
If you cannot fill a cohort of eight, the recorded course will not sell either. That is unwelcome information delivered cheaply.
What AI changes about course economics
Production is cheaper: outlines, slide copy, exercise variants, captions, transcripts and worksheets all come faster. So does everyone else's, which is why generic introductory courses on common topics have collapsed in price.
What holds value is what a model cannot supply: your specific experience, real examples with real numbers, feedback on student work and the accountability of a cohort.
Design the offer around those. Content-only products compete with free; access and feedback do not.
Frequently asked questions
How much should I charge for a course?
Price against the outcome's value to the student, not against course-length norms. Professional skills with clear career value support far higher prices than hobby topics.
Self-paced or cohort?
Start cohort-based for validation and revenue, then convert to self-paced once the structure is stable. Many creators keep both at different prices.
Do I need a large audience first?
You need buyers, not followers. A hundred people with the problem and budget is enough for a profitable first cohort.
Can AI create the course for me?
It can produce the scaffolding and exercises. If it produces the teaching, students are paying you for something they can obtain for free.
Which platform should I use?
Whichever gets you selling this month. Platform choice affects margin at the edges; audience decides whether there is a margin at all.
Should courses be evergreen or cohort-based?
Cohorts price higher and finish more often. Evergreen scales but completion rates fall sharply, which affects testimonials and referrals.