Author: Josh Kroman, Chief Operating Officer
By the time a finished course arrives for translation, most of what it will cost has already been set. Not by the word count, and not by the number of languages. By decisions a designer made months earlier, usually without knowing a second language was coming: where the text lives, how the audio was timed, how the quiz was built, which fonts were chosen.
The translation itself is the cheap part. What costs money is rebuilding a course that was never meant to hold another language.
On a typical course, translating the words is a minority of the invoice. The rest sits in four places:
Rebuilding anything with text inside it. Diagrams, screenshots, buttons and infographics with labels baked into the image have to be recreated once per language, by someone with the source file. If the source file is gone, they have to be recreated from scratch.
Nine choices do most of the damage. All nine are free to avoid at storyboard and expensive to undo later.
Text inside graphics
A process diagram exported as a flat image is now a design job in every language, not a translation job. Keeping labels as live text over the graphic, or keeping the layered source file, removes that cost entirely.
On-screen text tied to narration timing
When a caption appears at 00:14 because that is where the English narrator says the word, every language needs that cue moved. Designing the slide so text appears on a user action, or on a generous interval, means the timing survives.
Fixed-width text containers
A button sized exactly to the word "Submit" in English will not hold the same word in German or Finnish. Leaving roughly a third more room than the English copy needs is the cheapest insurance in the whole process.
Examples that only work in one country
Names, currency, addresses, holidays, a reference to a specific regulator. These are not wrong, they are just untranslatable without a rewrite. If the course is likely to travel, the storyboard can choose examples that carry.
Quizzes ordered by letter
Distractors arranged alphabetically, or feedback that refers to "option C," break the moment the answers are translated and re-sorted. Randomized or content-referenced feedback does not.
Fonts without the characters you will need
A brand font with no Cyrillic, Greek, Arabic or CJK coverage forces a substitution in exactly the languages where a substitution is most visible. Checking glyph coverage during design costs minutes.
Subtitles burned into video
Burned-in text means re-rendering the video per language. A separate caption file means uploading a new one.
Concatenated strings
Building a sentence out of fragments joined at runtime works in English and produces nonsense in most other grammars. Full sentences with variables, rather than sentence pieces, avoids it.
Terminology nobody thought to ask for
The organization’s approved wording usually exists before the course does, somewhere. Asking for it at storyboard costs one email. Finding out it existed after the translation is finished costs a full revision pass.
None of this requires designing for languages you may never build. It requires a short list of habits:
Text stays as text, in a layer or a variable, never flattened into an image.
Layouts leave room for copy to grow.
The narration script lives in its own file, matched to slide IDs.
Interaction feedback sits in a text layer rather than inside a trigger.
Source files for every graphic are kept and named.
Video carries a caption track rather than burned-in text.
Fonts are checked against the scripts the organization actually operates in.
A recent multi-session course went into Spanish without the client’s preferred terminology. Nobody withheld it. It existed, and the question of sharing it never came up before the course was built.
The translation finished on time. Then the client reviewed it, found their own terms missing, and the full course went back for revision. By that point the course was already loaded, so the corrections had to be made inside the learning platform rather than in the translation tool, which is slower and has to be done session by session. Every session needed a second review pass. Delivery moved by about a week.
A term list shared before the storyboard was signed off would have cost one email.
When the question comes up at handoff, there are only two answers, and neither is good. Re-engineer the course, which costs real money and delays the launch in every market. Or accept a weaker version in the other languages: cropped text, an English diagram sitting in a Japanese module, a quiz that no longer makes sense. Teams usually pick the second, then conclude that localized courses are lower quality by nature. They are not. They inherited a design that only had room for one language.
Do we have to decide the target languages at storyboard? No. The habits above cost nothing if the course is never translated, and they are what make translation affordable if it is. The decision that matters is whether another language is possible, not which one.
Does machine translation change this? It changes the price of the words and nothing else. Machine translation does not rebuild a graphic, re-time an animation, or find feedback text buried in a trigger. On a course that was not designed for it, the engineering is the majority of the work either way.
Voiceover or subtitles? It depends on how the course is used rather than on budget alone. Subtitles are cheaper and fine for desk-based learning. Voiceover matters where learners are doing something with their hands, or where reading speed in the target language is a barrier. Either way, the decision affects how the slide should be timed, which is a storyboard question.
What if the course already exists? Then the first pass is an audit rather than a translation: find the flattened graphics, the hard-coded strings and the timed builds, and price the fixes once rather than once per language.
The cheapest moment to make a course work in another language is the moment before anyone has built anything. The Global eLearning Launch Checklist covers the storyboard-stage items above in a form you can hand to a designer.
If you are looking at a library that was never designed to travel and want to know what it would take, Talk to a Specialist.