Most HR teams frame multilingual onboarding as a choice between subtitles and dubbing, as if one approach had to win. In practice, the more useful question is which video needs which treatment. Subtitles work well as the baseline for every onboarding video. AI dubbing pays off on a smaller set of modules where new hires have to watch and listen at the same time. This guide compares both approaches, gives HR and L&D teams a simple decision matrix, and covers a question that is often left to IT: where the videos and the viewing data actually live.
1. Why Onboarding Videos Need More Than One Language
In distributed organizations, new hires often do not share a first language: a warehouse team with several nationalities, a sales office abroad, remote hires in another country, or seasonal staff who join for a few months.
Video already carries much of a typical onboarding program, from the pre-arrival welcome message to tool walkthroughs and mandatory training. If you are still mapping those stages, our guide to creating a successful onboarding program breaks them down from pre-arrival to the first month.
Language matters most in three places:
- Comprehension in the first weeks. New hires absorb a lot of information at once. Doing that in a second language adds effort and increases the chance that something important is misunderstood.
- Mandatory and safety training. A completion record proves that a video was played. It does not prove that the content was understood. For safety and compliance topics, that difference matters.
- Belonging. Hearing the company’s message in your own language signals that the organization has thought about you before day one.
2. Option One: Subtitles
Subtitles are the translated text of what is said, shown on screen. For most organizations they are the natural starting point.
Strengths
- Low cost and fast to produce – Once a reviewed transcript exists, adding another language is mainly a translation and review task.
- Accessible – Captions help employees who are deaf or hard of hearing and anyone watching in a noisy environment or with the sound off.
- Searchable – The transcript can be reused in a knowledge base and for quizzes.
- Easy to update – When a policy changes, a subtitle file can be corrected without touching the video.
Limits
- Reading load – Viewers have to split their attention between the text and the image. That works for a talking head, much less for a demonstration.
- Hands-on and safety content – Someone learning to operate a machine or follow a lockout procedure needs their eyes on the action, not on the bottom of the screen.
- Mobile viewing – On a phone, subtitles cover a large share of the screen.
- Reading confidence – Employees who read slowly in the target language, or who are more comfortable listening than reading, get less out of a subtitled video.
For production work, one mistake comes up again and again: subtitles burned permanently into the video file. Every later correction then means a new export and a new upload in every language. Separate subtitle files, for example in SRT or WebVTT format, can be fixed in minutes.
3. Option Two: AI Dubbing and Voice-Over
There are two ways to replace reading with listening. A voice-over adds a translated audio track while the original voice stays faintly audible underneath. Dubbing replaces the original voice completely. With AI dubbing, that new voice is generated synthetically from a translated script.
Strengths
- Attention stays on the screen – New hires can follow a demonstration, a software walkthrough or a site tour without reading at the same time.
- Frontline and practical training – For shift workers, technicians and warehouse staff, a dubbed video is often the only format that fits how they actually learn on the job.
- A more personal experience – A welcome video in the viewer’s own language feels closer than the same video with text underneath.
Limits
- Review effort – An AI voice reads exactly what is in the script. Errors in translation, pronunciation of product names or tone have to be caught by a human reviewer before release.
- Company terminology – Internal job titles, abbreviations and product names need a glossary, or the dubbed version will sound generic or wrong.
- Timing – Translations are often longer than the source, which can push speech out of sync with what happens on screen. Scripts sometimes need to be shortened, not just translated.
- Consent for voices. If a real person’s voice is cloned, for example the CEO in a welcome video, that person needs to agree in writing, including how long the voice may be used.
- Transparency – Viewers should know when a voice is synthetic. Article 50 of the EU AI Act sets transparency obligations for AI-generated audio and video content, with disclosure duties for deepfakes, and a cloned voice of a real colleague can fall into that category. A short note at the start of the video is a simple way to be open about it.
4. A Simple Decision Guide
The choice becomes easier when you look at each video separately. These five criteria cover most cases:
A practical way to use the matrix: go through your onboarding videos one by one. If a video lands in the right-hand column for two or more criteria, it is a candidate for dubbing. Everything else gets subtitles.
5. The Hybrid Approach
For most organizations, the answer is not either-or. A hybrid setup works well:
- Subtitles for every video and every language. This is the baseline that keeps the whole library accessible and searchable.
- Dubbing for the modules with the highest impact. Typically safety instructions, the essentials of first day and the core processes of a role.
- One reviewed script as the single source. Start with a clean transcript, agree on the glossary, translate and review once. Subtitles and dubbed audio are then produced from the same approved text, so both versions say the same thing.
Video does not replace the people side of onboarding. A buddy, a manager check-in or a short team session still does work that no recording can, a point HR Shelf makes well in its tips on how to make your employee onboarding process engaging. Multilingual video takes repetitive explanations off the trainers’ plate, so those conversations have more room.
The same rule works beyond onboarding, for example in corporate training programs subtitles are required everywhere, and dubbing where attention and comprehension matter most.
6. Where the Videos Live: GDPR-Compliant Hosting
Onboarding videos often show employees, internal processes and sometimes floor plans or systems that should not be public. The viewing data is sensitive too: who watched which module, when and how far is personal data. For HR and L&D teams in the EU, and for anyone onboarding EU employees, a few questions belong on the checklist before a video goes live:
- Where are the videos and the viewing data stored and processed? Ask about the hosting location, the content delivery network and any subprocessors.
- Is a data processing agreement in place? Under Article 28 of the GDPR, a provider that processes personal data on your behalf needs a written agreement with you.
- Does the player set tracking cookies without consent? This is a common issue with free video platforms, where the embedded player may load third-party tracking before anyone presses play.
- How is access limited to employees? Single sign-on, private links and settings that keep videos out of public search results are worth checking.
- What happens to the data later? Define how long viewing data is kept and how videos and data are removed when they are replaced or when employees leave.
- Who needs to be involved? In some countries, such as Germany, employee representatives have a say when a tool can be used to monitor behavior or performance, and detailed viewing analytics can fall under that.
For teams that want to go deeper, the guide to GDPR-compliant video hosting for companies covers data sovereignty, content protection and evaluation criteria in more detail.
7. Checklist Before Rollout
- Glossary. List company terms, product names, job titles and abbreviations, and decide which ones stay in English.
- Native-speaker review. Have subtitles and dubbing scripts checked by people who speak the target language natively, ideally colleagues from the local team.
- On-screen text. Translate or explain text in slides, graphics and software screenshots. Subtitles do not help if the key instruction is written in the image.
- Accessibility. Captions should include relevant sounds such as alarms or machine signals, and contrast and font size should work on small screens.
- Completion and comprehension. Track completion, but add a short quiz or a single question such as “Was anything unclear?” in every language. If results in one language fall behind the others, check the translation before blaming the learners.
- Update ownership. Decide who updates which language version when the source video changes, and how quickly.
Conclusion
Subtitles or AI dubbing is the wrong question when it is asked once for the whole program. Asked video by video, the answer is clear: subtitles as the accessible baseline, AI dubbing where new hires must watch and listen at once. Add a shared glossary, a native-speaker review and hosting that respects employees’ data, and multilingual onboarding becomes routine rather than a special project.
