Author and editorial responsibility
Tim Jamboula, Founder of Corporathon. Last reviewed 24 August 2026.
Summary
An online course is cheap, flexible and good for basics, but weak at transferring to your real process. Inhouse AI training costs more per head but aligns to your data and roles. When a concrete process should run better afterwards, a hands-on format on your real case beats both. The choice depends on goal, data relevance and completion rate, not on list price.
Table of contents
- The real question behind the format choice
- What each format actually delivers
- The decision framework in four questions
- The hidden cost logic of completion rate
- A worked ROI example (model)
- Why transfer beats format
- EU AI Act: what counts
- What to do next
- FAQ
1. The real question behind the format choice
"Inhouse or online" is usually asked as a cost question. The better question: how much transfer to your real process does the result need? An online course scales knowledge cheaply and broadly, inhouse training brings your role and data context, but neither automatically ends in application. That gap decides the return.
The cheapest course is expensive if no one finishes it and no one applies what they learned. – Tim Jamboula, Founder of Corporathon
2. What each format actually delivers
| Criterion | Online course | Inhouse AI training | Hands-on hackathon |
|---|---|---|---|
| Cost per head | low | medium to high | outcome-based, not per head |
| Relevance to your data | none | partial, on examples | direct, on real processes |
| Transfer to daily work | low without follow-up | medium | high, because you build immediately |
| Completion rate | often low | high, because scheduled | very high, output is required |
| Output | certificate | certificate plus context | working prototype |
| Scales for breadth | very well | limited | focused, not for mass |
All three are legitimate. The online course is strong for broad basics, inhouse training for context-near knowledge, the hackathon for visible behaviour change on a concrete case.
3. The decision framework in four questions
- Breadth or depth? Broad basics for many favour the online course. Depth on a process favours inhouse or hackathon.
- Does the result need your real data? If yes, the pure online course is out.
- What is the realistic completion rate? If self-paced courses rarely get finished, a scheduled format is more honest.
- Should there be a usable result at the end? If yes, the hackathon is the direct path.
Rule of thumb: breadth without data relevance means online course. Depth with data relevance and a required result means inhouse or hackathon.
Find the format that actually lands with your team
Two paths, depending on your goal.
- Book directly: Book a discovery call. 30 minutes, we place online, in-house and hackathon against your goal.
- Read first: Enter your email and get the format decision matrix. No spam, unsubscribe any time.
4. The hidden cost logic of completion rate
The list price of an online course is misleading because it ignores the completion rate. Factor it in and the comparison often flips.
- Effective cost per completion. A cheap course with a low completion rate costs more per person who actually learned than it appears to.
- Opportunity cost. Self-paced time without application is work time without return.
- Context cost. Missing data relevance creates rework to translate what was learned to your own case.
- Result cost. No course ships a prototype, you need that separately.
We quote prices personally. The right scope emerges in conversation, from these variables.
5. A worked ROI example (model)
Illustrative model: fifty people get a 200 EUR online course, 10,000 EUR list cost. At a 30 percent completion rate, 15 people effectively learn, about 667 EUR per learner. A scheduled inhouse format at 90 percent completion teaches 45 people, cutting the effective per-head cost sharply. A hackathon is measured differently: not per head, but against the value of the built prototype (see the model in our cost article). The figures are not a guarantee and not client numbers. They show that the completion rate is the real cost metric.
6. Why transfer beats format
Visualisation: knowledge against application by format (schematic, illustration without measured values).
Application
high | ● Hackathon
|
|
medium| ● In-house
|
|
low | ● Online course
+------------------------------------------> Breadth of knowledge
narrow/deep broad
The point is not that one format wins, but that breadth and application rarely sit in the same format. If you want application, do not buy a pure knowledge format.
7. EU AI Act: what counts
Since 2 February 2025, Article 4 requires sufficient, role-based AI literacy. An online course documents attendance, inhouse adds context, a hackathon documents application. Each is a building block, none is an official certificate, none guarantees automatic compliance. The company itself must assess whether this is adequate.
8. What to do next
For broad basics without data relevance, an online course is the cheap choice; watch the completion rate. For context-near knowledge on your roles, inhouse fits. When a concrete process should run better afterwards, the direct path is a hands-on hackathon, one week from first contact to prototype.
Next step
- Book directly: Book a discovery call. We place your format against the real goal.
- Read first: Enter your email and get the format decision matrix. Double opt-in, unsubscribe any time.
Related terms
AI hackathon · AI training vs AI hackathon · AI training cost · AI adoption
FAQ
Is inhouse training always better than an online course? No. For broad basics without data relevance an online course is cheaper and enough. Inhouse pays off when context, roles and real data matter, or when self-paced completion rates are low.
Why does the completion rate matter so much? It determines the effective cost per learner. A cheap course only a third finish often costs more per head than a scheduled format with high completion.
When is a hackathon worth it instead of both? When a concrete process should measurably improve afterwards and an owner carries the result forward. Then the hackathon delivers knowledge, application and an artifact in one step.
Is an online course enough for the EU AI Act? It can document attendance but is only a building block. Article 4 turns on the role- and context-appropriate adequacy of the overall program, which the company owns.
Legal note: Online course, inhouse training and hackathon can document AI literacy measures, but are not an officially mandated certificate and do not automatically guarantee compliance with Article 4.