Subjects
The best online courses by subject, and which platform teaches each one properly
No platform is best at everything. These guides pick the subject first and then the platform, which is the order that produces better decisions.
Choosing a platform first and a subject second is how people end up learning statistics somewhere that teaches illustration well. Each guide below starts from the subject and works outwards to whichever platform teaches it seriously, which occasionally means recommending something other than a course at all.
- Are Coursera Certificates Worth It? What Employers Thinkcoursera
- Best MasterClass Coursesmasterclass
- Best Coursera Coursescoursera
- Best Online English Coursesmulti-platform
- Best Skillshare Classesskillshare
- Best YouTube Coursesmulti-platform
- Is Skillshare Free? What You Get Without Paying Anythingskillshare
- Skillshare Free Coursesskillshare
- What Is MasterClass? The Format, the Price and the Pointmasterclass
Where each platform is genuinely strongest
| Subject | Where it is taught best |
|---|---|
| Data, statistics, machine learning | Coursera and edX, for graded work and university teaching |
| A specific tool or framework | Udemy, bought outright on sale |
| Business software and workplace skills | LinkedIn Learning, especially through a library card |
| Illustration, lettering, design projects | Skillshare, judged by the class project |
| Photography and the business around it | CreativeLive, in long workshop form |
| Writing craft and creative process | MasterClass, as insight rather than instruction |
| History, philosophy, science for the general reader | The Great Courses, as lectures |
| Speaking a language | One-to-one tutoring, which no course platform replaces |
The three kinds of subject
Subjects that need marking
Statistics, accounting, anything with a right answer you can get wrong without noticing. These belong on platforms that grade work, which in practice means the university platforms and, for technology, the ones with real assessments. A cheaper subscription that never tells you whether you are wrong is not a saving in these subjects.
Subjects that need repetition
Illustration, lettering, cooking, photography, most software. Progress comes from doing the thing many times, so the platform’s job is to define a project and get out of the way. Short project-based classes beat long lectures here, and the certificate is irrelevant.
Subjects that need a person
Speaking a language, professional writing, presenting, negotiating. No recorded catalog can tell you that your sentence is grammatical and still wrong. The efficient approach is to take the input free or cheap and spend the money on hours with somebody who responds.
Why platform-first choosing goes wrong
The usual sequence is backwards. Somebody reads that a platform is good, subscribes, and then looks for their subject inside it. Because every catalog here is large, they will find something, and because the platform was chosen for reasons unrelated to that subject, what they find is frequently the weakest part of the catalog.
Choosing subject-first inverts that and takes about the same amount of time. It also produces the occasional recommendation that no platform is right, which is a genuinely useful outcome and one a platform-first search can never reach. Two of the guides here end that way, and both are more valuable for it.
The other advantage is cost. Subject-first choosing frequently ends in buying one course rather than subscribing for a year, because the specific thing you needed exists on a per-course platform and does not require access to nine thousand other courses to reach.
Two subjects worth a warning
Data science is the most oversold subject in online learning. Every platform has a path, the paths look alike, and the difference between them is whether the statistics underneath is taught or assumed. Tool fluency without it produces somebody who can manipulate a dataset and cannot say when a result is meaningless, which is the failure mode employers in the field complain about most. Take the math somewhere that marks it, even if the tooling comes from somewhere cheaper.
Business and leadership is the second. The format rewards anecdote, the anecdotes are retrospective, and very little of what is taught can be tested against anything. There are exceptions where an instructor has an actual method, and they are worth finding, but the category average is the weakest on every platform in this comparison and the pricing does not reflect that.
The prerequisite problem
Most abandonment in technical subjects happens because a prerequisite was missing rather than because the course was bad. A data course assumes statistics, a machine learning course assumes linear algebra, and neither says so loudly enough for a beginner to notice before paying.
Fixing the foundation is free on Khan Academy and costs two weeks. Doing that first is the single highest-return decision in this whole market, and it is the one almost nobody makes.
Where a subject sits changes the platform
The same subject can belong to two different categories depending on the level. Beginner photography is repetition: shoot, review, shoot again, and a project platform serves it well. Advanced photography is judgment, and judgment improves when somebody whose work you respect tells you why an image does not hold. That is a different purchase, and no subscription supplies it.
The practical implication is that a platform recommendation has a shelf life. The place that taught you the fundamentals of a craft is rarely the place that takes you past competent, and staying subscribed out of habit is the most common way this market quietly stops paying off.
The rule behind the table
Subjects that need marking belong on platforms that mark. Subjects that need repetition belong on platforms built around projects. Subjects that need a person listening to you belong with a person. Most disappointment with online learning comes from ignoring which of the three a subject is, and no amount of catalog size fixes that mismatch.
Questions people actually ask
Which platform is best for learning a language?
None of them, for speaking. Recorded courses cover grammar, listening and reading well, and fluency comes from producing language that somebody corrects, which means one-to-one tutoring rather than a subscription.
Where should you learn programming?
Foundations on a university platform or free on Khan Academy, a specific language or framework on Udemy bought outright, and an ordered path through a stack on Pluralsight if an employer is paying.
Are creative subjects better served by specialist platforms?
Yes. Skillshare, CreativeLive and Domestika teach illustration, design and photography better than the general platforms, and none of them certifies anything, which for a portfolio subject does not matter.
Which subjects should you not learn from video at all?
Anything whose whole point is production judged by a person: writing at a professional level, speaking a language, presenting. Video teaches the input side and cannot mark the output.