The real cost of translating your docs
Everyone talks about translation costs in the abstract. Let's use real numbers.
Take a concrete scenario: a documentation site with 200 pages, translated into 5 languages (Spanish, French, German, Japanese, Chinese). This is roughly the size and scope of projects like Prettier or Jest — both of which we've translated as live demos.
Initial translation cost
Human translation agencies
Professional technical translation runs $0.10–0.30 per word (industry benchmarks from Nimdzi and GALA range $0.09–0.35 depending on language pair and specialization). Technical documentation averages around 500 words per page.
- 200 pages × 500 words = 100,000 words
- 100,000 words × $0.15/word = $15,000 per language
- 5 languages = $75,000
Timeline: 4–8 weeks per language, partially parallelizable. Expect 6–10 weeks end-to-end.
This doesn't include project management, reviewer coordination, or the formatting work to get translations back into your docs framework.
Generic machine translation (Google Translate, DeepL)
API costs are low — typically $20 per million characters.
- 200 pages ≈ 500,000 characters
- 500,000 × $0.00002 = ~$10 per language
- 5 languages = ~$50
Sounds great. But:
- Someone needs to post-process every page to fix code blocks, inline code, and component structure
- Someone needs to manually ensure terminology consistency
- Someone needs to re-insert translations into your docs framework with the correct file structure, frontmatter, and sidebar config
- Someone needs to deploy it
The API cost is the smallest part. The engineering time to make the output usable is the expensive part — easily 40–80 hours of work, worth $4,000–$12,000 at typical rates.
Effective total: $4,000–$12,000 with significant quality compromises.
PageTurner
$0.05–0.15 per page per language, depending on page complexity and content density.
- 200 pages × $0.10/page = $20 per language
- 5 languages = ~$100
This includes: terminology extraction, consistency enforcement, code/component preservation, framework-aware file generation, and deployment to Vercel.
Timeline: 1–2 hours for 200 pages across 5 languages.
The part people forget: maintenance
Initial translation is a one-time cost. Maintenance is forever.
Documentation changes with every release. A typical active project updates 10–30% of its docs per quarter. For our 200-page example, that's 20–60 pages changing every few months.
Human translation maintenance
Every update cycle requires:
- Identify which pages changed
- Send changed pages to translators
- Wait 1–2 weeks for turnaround
- Review translations
- Reformat and redeploy
Cost per update: $1,500–$4,500 (20–60 pages × 5 languages × $0.15/word × 500 words).
Annual cost (4 updates): $6,000–$18,000.
This is why most human-translated documentation falls out of date. The ongoing cost is unsustainable for all but the largest companies.
Generic MT maintenance
The API cost per update is negligible (~$5–10). But the engineering time repeats: fix code blocks, verify consistency, reformat, redeploy.
Even if the initial setup created scripts to automate some of this, each update still requires 8–16 hours of engineering time.
Annual cost: $3,200–$9,600 in engineering time.
PageTurner maintenance
Updates cost 60–80% less than initial translation because only changed content gets retranslated. Translation memory preserves unchanged content.
- 40 changed pages × $0.10/page × 0.3 (change ratio discount) × 5 languages = ~$6 per update
- Detection, retranslation, and redeployment are automatic — no manual work
Annual cost: ~$25 for 4 update cycles. Zero engineering time.
Total cost over 2 years
| Human Translation | Generic MT + Engineering | PageTurner | |
|---|---|---|---|
| Initial | $75,000 | $4,000–12,000 | ~$100 |
| Year 1 maintenance | $6,000–18,000 | $3,200–9,600 | ~$25 |
| Year 2 maintenance | $6,000–18,000 | $3,200–9,600 | ~$25 |
| 2-year total | $87,000–111,000 | $10,400–31,200 | ~$150 |
| Quality | Excellent (95–98/100) | Poor without post-editing (65–70/100) | Strong (91/100) |
| Time to first deploy | 6–10 weeks | 1–2 weeks (with eng work) | 1–2 hours |
| Maintenance effort | Manual per update | Semi-manual | Automatic |
These numbers reflect the initial investment. But the return on localization is well-documented: a Forrester Total Economic Impact study found that modern translation workflows deliver 345% ROI with 90% time reduction. Nimdzi Insights reports that 96% of companies see positive ROI from localization, with 65% achieving 3x return or more.
When each option makes sense
Human translation makes sense when you need perfect, publication-quality output and have the budget — typically enterprise companies with $10M+ ARR translating customer-facing documentation into 2–3 strategic languages.
Generic MT + engineering makes sense for internal documentation or low-stakes content where "good enough" quality is acceptable and you have engineering capacity to build and maintain the pipeline.
Automated translation with quality enforcement (what PageTurner does) makes sense for developer documentation where you need high quality across many languages, fast turnaround, and sustainable maintenance — without dedicating engineering resources to a translation pipeline.
One thing these numbers don't capture: the cost of waiting. Lingoport estimates that retrofitting internationalization into an established product costs $500K–$2M. The earlier you start, the less it costs.
See the output
Don't take our word for the quality. Browse the live demos:
For a detailed estimate for your documentation, contact us. For open source projects, apply to the Founding Partner Program for free translation.