A product tour is the most reusable asset a SaaS company owns and the one most often rebuilt from scratch. Five markets, five screen recordings, five scripts and a marketing team that now maintains five divergent videos.
It does not have to work that way, but the reason it usually does is decided before anyone opens a recording tool.
What makes a demo expensive to localise
Three things, in order of cost:
On-screen UI text. If the recording shows dense interface copy, every language needs a new recording or a compositing pass over moving UI. This is the single biggest driver.
Voice-over tied to screen action. A script that says "and here, as you can see, in the third panel" locks the narration to the timing of a specific screen.
Idiomatic copy. Marketing lines that lean on English wordplay cannot be translated; they have to be rewritten, which means a writer per market rather than a translator.
None of those are inevitable. They are decisions made at script stage by people who were not thinking about market six.
The design rules that make five languages cheap
Record fewer, longer screens. A tour that moves through six stable screens localises easily. One that scrolls through twelve panels does not.
Keep on-screen overlays as a separate layer. If the text sits above the recording rather than inside it, the recording is language-neutral and only the overlay changes.
Write to the beat, not to the second. Narration that lands on transitions survives re-voicing. Narration that lands on a specific frame number does not.
Avoid English wordplay in the claim. The claim should be a proposition, not a pun. Puns are untranslatable and the market will notice the gap.
Decide the language set at brief stage. Five languages chosen up front is one production. Five added over a year is five.
The workflow that works
| Stage | Cost profile |
|---|---|
| Script, one language, market-neutral | Full production, once |
| Product recording | Once, language-neutral |
| Edit, motion, captions structure | Once |
| Voice per language | Language layer |
| On-screen text per language | Language layer |
| Native review per market | Per market, non-negotiable |
The bottom two rows are where quality is won or lost. A localised demo fails not because the voice sounds synthetic but because a native speaker hears a sentence no local marketer would write.
What to rewrite rather than translate
Four parts of a demo script are always rewritten, never translated:
1. The opening claim. Different markets tolerate different levels of assertion. A headline that reads confident in New York reads overstated in Frankfurt.
2. The objection handling. The objection itself differs — data residency in the EU, procurement cycles in the Gulf, integration ecosystems in Japan.
3. The call to action. "Book a demo" is not a universal conversion event. Some markets expect a trial, some a call, some a local reseller.
4. Numbers and units. Formats, currencies and date conventions, on screen as well as in voice.
The product screens themselves stay. That is the whole saving.
Order of operations, for a first run
If you are localising for the first time, sequence it so mistakes are cheap:
Market one, full production. Pick the market you understand best. Get the script and the beat structure right in one place before multiplying it.
Markets two and three, as a layer. Same edit, new voice and overlays. If something does not fit — a claim that will not land, an overlay that runs long — you learn it here at twice the price of one market rather than five times.
Markets four and five, once the pattern holds. By now the overlay template and the review loop are established and the marginal cost is at its floor.
Doing all five at once is defensible only if you have already localised a different asset into those languages and know the reviewers.
Where it goes wrong in practice
A translator instead of a writer. Translation preserves the source structure. A demo for a market needs the local structure.
Review by a fluent speaker from the wrong region. European Spanish reviewed by a Mexican speaker is not wrong, it is off, and off is what prospects notice.
Overlaying long text into a fixed frame. German and Dutch run longer than English in the same width. Either shorten the copy or rebuild the overlay; do not shrink the type until it is legible only on a desktop.
Re-recording the product "properly". This is the expensive reflex. Nine times out of ten the recording is fine and the overlay is the problem.
The maths
For a five-market demo, one production plus four language layers lands at roughly the price of two new assets rather than five. That ratio is why localisation belongs in the plan instead of the change order — and why a SaaS team running a quarterly release cadence can ship a multilingual tour every quarter without a second production team.
Where NNI sits
We produce SaaS and technology video at from $100 a finished asset — product tours, explainers, feature and changelog video — and language variants as a layer from a single master, with native review, delivered in 24–48 hours per batch. Under your brand, no supplier marks.
If your demo is currently recorded in one language and subtitled in four, the subtitle is not a localisation strategy. It is a caption.
---
Jayesh runs NNI Design Studio, white-label video production for agencies and SaaS teams. Rate card at nnidesignstudio.com/pricing; five-language quarterly volume on Always-On.
Drop your take 0
// no hot takes yet. start the thread 👇