The question comes up first almost every time. And it is hard to answer, because the spread is absurd: the same ten-minute video costs anywhere between a few hundred euros and a five-figure sum depending on the route.
So here is no single number, but the structure behind it. Once you know that, you can place any quote that lands on your desk.
The market benchmark to orient by
For one hour of produced eLearning, external production costs run up to 40,000 euros. That is a market benchmark and an upper bound, not an average. It applies to elaborate productions with a film crew, voice talent, editing, interactions and several rounds of review.
The figure is still useful because it gives you a reference. Calculate your own projects per finished learning minute, not per project. Only then can quotes be compared.
The three routes and what they cost
Route 1: agency or production partner
You buy an outcome and accountability. Pricing follows shoot days, editing effort, voice hours and correction cycles. Quality is high, so is the price per minute. Two things are critical: changes cost almost as much as the original production, and every language version is a separate quote.
Route 2: in-house production with a classic authoring tool
The licence is manageable, the staff cost is not. Be realistic: concept, script, alignment with the business unit, recording, editing, voiceover, quiz, LMS import. In many teams the effort per learning video runs to about a working day, and more for software topics with a screencast. Multiplied by an internal day rate that is the real cost block, it just never appears on an invoice.
Route 3: AI-supported production
The licence is the visible line item, the effort per video drops considerably. A solid reference value from a named, approved customer: at DPD Germany production time per learning video fell from about a working day to around an hour. Measured across more than 1,800 videos in two and a half years, produced by roughly 240 active authors.
The line items missing from almost every calculation
This is where it gets decisive. The initial production is rarely the expensive part.
- Maintenance and updates. Every software release, every changed work instruction, every new legal requirement makes content stale. With classic production that means recording again, editing again, importing into the LMS again.
- Translation. Per language a separate recording, a separate review, a separate approval.
- Review loops. The item nobody budgets and everybody knows. Between first script and sign-off there are often three to five rounds.
- Opportunity cost of the backlog. If a safety briefing takes six weeks, six weeks run without a completion record.
- LMS handwork. Import, versioning, assignment, follow-up. Sounds small, adds up across dozens of courses.
Where the maths tips over
As long as you produce five videos a year, the route hardly matters. An agency is often the cheapest option then, because you build no internal capacity.
Beyond a certain volume that reverses. The tipping point usually sits where three things meet: more than roughly 20 videos per year, at least two languages, and content that changes regularly. From there, maintenance and translation dominate total cost, and those two items are exactly the ones automation reduces most.
A concrete example of the mechanism: 140 languages for subtitles and translation, 50+ languages for voiceover, 120+ avatars. A language version is no longer a project, it is an export. And SCORM remote update means a content correction does not have to travel through the entire LMS process.
How to calculate your own case
Take your last ten produced trainings and pull together four values:
- Finished learning minutes in total
- Internal person-hours, honestly including alignment
- External cost for voice, editing, translation
- Number of updates in the last twelve months
Divide items two to four by the first. You now have your real price per learning minute. Almost every team doing this for the first time finds they had underestimated it, because internal hours were never tracked.
The most common wrong decision
Teams compare licence prices. That is the smallest variable in the equation.
What matters is cost per update and cost per language version. Optimising those two lowers total cost far more than negotiating the licence.
In your next vendor conversation ask only two things: what does it cost when minute four changes? And what does the third language cost? The answers sort the market faster than any demo.
