Help center to course is the conversion with the biggest gap between how easy it looks and what it actually requires.
It looks trivial. You already have 80 articles, written by people who know the product, edited, screenshotted, kept current because support depends on them. Surely that’s a course with a different stylesheet.
It isn’t, and the reason is in the reader’s state. A help article is written for someone who is already stuck. They arrived with a specific problem, searched, and want the answer in nine seconds. Customer education is written for someone who isn’t stuck yet, and its entire job is making sure they don’t get there.
Same facts, opposite premise. Here’s how to do the conversion properly.
Try Mini Course Generator free – turn the articles customers search for into a course they take first.
Key Takeaways
- Help articles answer “how do I do X”. Courses answer “what should I be doing, and in what order”. The reordering is the work, not the writing.
- Use your own analytics to pick the source articles. The most-viewed help pages tell you exactly where customers struggle, and that ranking should decide the course’s contents.
- Don’t convert the whole help center. 80 articles becomes 6 to 8 lessons, and the ratio is the point.
- Keep the help center. This is an addition, not a migration – the articles still have to answer the stuck customer, and they’re probably how customers find you when something breaks.
- Where the course is delivered matters as much as what’s in it. In-product beats a portal, because the customer is already there.
Reactive Content vs Proactive Content
Worth making the distinction sharp, because it decides every editing choice that follows.
A help article assumes a question. Its title is the question. It gets to the answer fast, covers the edge cases, and links sideways to related articles. It’s a good help article precisely because it doesn’t set context – the reader supplied the context by arriving.
A course assumes no question. Nobody arrives at lesson one with a specific problem. So it has to establish why this matters before explaining how it works, order things so each builds on the last, and check understanding rather than assume it.
Converting between the two isn’t a rewrite. It’s a reordering plus an addition: sequence the facts by how a customer encounters them rather than by how they’re searched for, and add the “why this matters” that a help article correctly leaves out.
That’s why the naive approach – one lesson per article – produces something that reads like a manual with progress bars.
Let Your Analytics Choose the Source
The best input here isn’t the article list. It’s the view counts.
“From the help center, list the 20 most-viewed articles in the last 6 months with their view counts. Group them into themes.”
Your most-viewed help articles are a ranked map of where customers get stuck. That ranking should decide what the course covers, and it usually disagrees with what the product team assumes.
The two useful patterns:
A high-traffic article about something basic means the product or the onboarding isn’t communicating it. That’s lesson one, not an appendix.
A cluster of related articles all getting traffic is a single missing concept, not five missing answers. That’s one lesson, and it’s usually the highest-value one in the course.
If support tickets are also visible, cross-reference. An article with high views and continuing tickets on the same topic means the article isn’t working, and rewriting it as a lesson without fixing the underlying explanation just moves the problem.
The Prompt
This assumes an assistant connected to your help center – or able to read its public help pages – and to a course platform over an MCP server. Setup is four steps, covered in the Claude connectors walkthrough.
“Using help center articles [list 12-15 by title], build a getting-started course in Mini Course Generator. Six lessons, ordered by what a new customer encounters first, not by article popularity. Each lesson should open with why this matters to the customer’s outcome before explaining the steps. Merge related articles into single lessons rather than one lesson per article. After each lesson add a quiz question that presents a realistic situation and asks what the customer should do. Publish it and give me the link.”
Three clauses carry the conversion:
“ordered by what a new customer encounters first, not by article popularity.” Popularity chose the contents; chronology has to choose the order. Those are different, and conflating them is the most common failure here.
“open with why this matters.” The sentence a help article correctly omits and a lesson can’t do without.
“merge related articles into single lessons.” The instruction that stops you shipping 15 thin lessons. Expect a 2-to-1 or 3-to-1 collapse.
Keep the Help Center
Worth stating plainly, because “we’re turning our docs into a course” sometimes gets heard as a migration.
The articles still have to exist. A stuck customer at 11pm wants the answer in nine seconds, not lesson four of a course. And your help center is very likely the first thing a customer with a problem finds, which is attention from people actively looking for what you do.
The course is a second surface for the same knowledge, aimed at a different moment: before the problem instead of during it. The two feed each other – the course reduces the tickets, and the tickets tell you what the next version of the course should cover.
If you’re building this into a wider programme rather than a one-off, the strategy side is in our guide to customer education.
Where the Course Should Live
This decides whether it gets taken, and it gets less attention than the content does.
| Delivery | Works when | The catch |
|---|---|---|
| In-product | The customer is already there, at the moment of relevance | Needs the platform to support embedding |
| Emailed link | Onboarding sequences, post-purchase | Competes with every other email |
| Inside your support messenger | The customer is already asking | Needs the messenger to support it |
| A customer academy portal | Deep, multi-course programmes | Another login, which is where most of these die |
The pattern across all four: the fewer new places a customer has to go, the more of them finish. A portal is the most impressive option and the one with the worst completion rate, because it asks for an account before it’s demonstrated any value.
For most B2B products the strongest option is in-product, which is what a headless LMS exists to do – the course renders inside your own interface rather than sending customers somewhere else.
What Needs Human Attention
Screenshots go stale faster than text. Help centers usually maintain them because support notices. A course, once published, has nobody noticing.
Edge cases don’t belong. Help articles are thorough about exceptions, correctly. A getting-started course covering every exception teaches a new customer that your product is complicated.
Cross-links break the sequence. Articles link sideways constantly. In a course, a link out is an exit, and a lesson full of them is a table of contents wearing a costume.
Tone shifts. Help writing is neutral and instructional. Customer education can be warmer, because the reader isn’t frustrated yet.
The Short Version
Help articles and customer education contain the same facts and assume opposite readers. One is written for a customer who’s already stuck; the other exists so they don’t get stuck. The conversion is a reordering plus an addition, not a rewrite.
Use your own view counts to decide what the course covers – your most-read help articles are a ranked map of where customers struggle – then order the lessons by what a customer encounters first, which is a different sequence entirely. Merge related articles rather than making one lesson each, and expect 80 articles to become 6 to 8 lessons.
Keep the help center. It answers the stuck customer and it’s probably where those customers first land. The course is a second surface aimed at an earlier moment.
Then put it where the customer already is. That choice affects completion more than the content does, and it’s the part customer training software has to get right before any of the rest matters.
Frequently Asked Questions
Can I turn my help center into a course automatically?
An assistant connected to your help center and a course platform can do the conversion in one request, but the instruction matters. Ask it to order lessons by what a customer encounters first and to merge related articles, or you get one thin lesson per article, in the order the articles were most viewed.
How many help articles make a course?
Fewer than you’d think. Around 12 to 15 selected articles typically become 6 to 8 lessons, because related articles collapse together. Converting an entire help center one-to-one produces a manual with a progress bar.
Should I delete the help articles after making the course?
No. They serve a different moment – a stuck customer wants an answer in seconds, not a lesson – and they’re likely the first thing a customer with a problem finds. The course is an additional surface, not a replacement.
Which help articles should the course cover?
The most-viewed ones, since view counts are a ranked map of where customers struggle. A high-traffic article about something basic belongs in lesson one. A cluster of related articles all getting traffic usually indicates one missing concept rather than several missing answers.
Where should customer education be delivered?
Wherever the customer already is. In-product delivery consistently outperforms a separate academy portal, because a portal asks for a new account before it has demonstrated value. Email works for onboarding sequences; a portal suits deep multi-course programmes.
What is the difference between a help center and customer education?
A help center is reactive: it answers a specific question for someone who already has a problem. Customer education is proactive: it’s sequenced, assessed, and designed so the problem doesn’t arise. Same knowledge, opposite assumptions about the reader.
