Short answer: Launch the 1–2 languages that already bring inquiries, use subdirectories (/en/, /ja/) with hreflang, and have native speakers review AI drafts. Adding two languages to a Chinese site costs about NT$120,000–350,000 over 6–9 weeks; a translation plugin with no review often brings fewer overseas inquiries, not more.
A 30-person metal-parts factory saw in GA4 that 40% of visits came from Japan and Southeast Asia, yet its site was Chinese-only. Overseas buyers read specs through browser auto-translate, and the company received just 6 English inquiries a year. Sales said "clients come from trade shows," but the website was turning people away.
When it fits and when it doesn't
Bottom line: do it when you have a clear overseas audience and the capacity to serve it; otherwise fix the Chinese site first.
| Good fit | Poor fit |
|---|---|
| Exporters or cross-border e-commerce with existing overseas inquiries | 95% of customers in Taiwan and no foreign-language sales staff |
| English or Japanese catalogs needed after trade shows | Just wanting to "look international" |
| Tourism, B&Bs and restaurants with many foreign guests | Heavy weekly content updates with nobody to translate |
| Overseas distributors or grants requiring foreign-language pages | Products restricted by local regulations |
Alternatives matrix
Bottom line: on a tight budget start with key English pages; a custom multilingual site suits companies chasing orders long-term.
- Browser auto-translate: free; not indexed by Google, and technical terms are often wrong.
- Translation plugins (e.g., WPML, Weglot for WordPress): about NT$3,000–25,000 a year, fast; unstable machine quality and limited URL control.
- 5–8 key English pages: about NT$30,000–80,000 for home, products, about and contact; good for testing a market.
- Custom multilingual site: about NT$120,000–350,000 with language fields, hreflang and inquiry routing; higher upfront cost.
The full process
Bottom line: it takes 6–9 weeks, and translation review, not coding, takes the longest.
- Week 1 | Language audit: rank languages using GA4 country data and inquiry records; deliver a priority list.
- Week 2 | Architecture: /en/ and /ja/ subdirectories, Figma notes on layout differences; deliver a page map.
- Weeks 3–5 | Translation and localization: DeepL or GPT draft, native review, and a glossary in Notion or Google Sheets.
- Weeks 4–7 | Development: multilingual fields in a Laravel back office, a language switcher, hreflang annotations, per-language sitemaps, and forms that notify the right sales rep.
- Weeks 8–9 | QA and launch: native-speaker acceptance and hreflang checks in Search Console.
Real costs
Bottom line: development is only half the bill; translation and ongoing updates are the long-term cost.
- Development (language architecture, back office, hreflang): NT$60,000–180,000.
- Translation review: about NT$1.5–3 per Chinese character; a 30-page site is roughly 20,000 characters, or NT$30,000–60,000 per language.
- Machine translation API: Google Cloud Translation is about US$20 per million characters, cheap, but still needs human review.
- Hidden costs: translating new pages (about NT$2,000–8,000 a month), foreign-language support staff, and per-language ad creatives.
Reality vs. expectations
Bottom line: launch is the starting line; inquiries take 3–6 months to show.
- Expectation: AI translation is ready to publish. Reality: one wrong spec, unit or certification name and buyers lose trust.
- Expectation: overseas inquiries next week. Reality: indexing and ranking new language pages often takes 3–6 months.
- Expectation: translate the homepage only. Reality: buyers mostly read spec pages and download catalogs.
Common traps and fixes
Bottom line: most failures come from URL structure and update workflow, not translation itself.
- Auto-redirecting by IP so Googlebot sees one version → use a manual switcher with a suggestion banner.
- hreflang tags that don't point back → every language page lists all versions plus x-default.
- Foreign-language form alerts going to the Chinese support inbox → route by language to the right rep.
- Chinese text baked into images → separate text from images or make one image per language.
- No glossary, inconsistent terms → fix translations for 50–100 key terms first.
Success metrics and a 90-day roadmap
Bottom line: watch indexing first, engagement second, inquiries last.
- Day 30: 80%+ of language pages indexed and zero hreflang errors.
- Day 60: foreign-page engagement time at least 70% of Chinese pages; rewrite the 5 pages with the highest bounce.
- Day 90: foreign-language inquiries and conversion rate versus the same period before launch; decide on a third language.
Decision checklist
Bottom line: if 7 or more of these 10 are "yes," a multilingual project is worth starting.
- ☐ We had overseas inquiries or orders in the past year
- ☐ GA4 shows 20%+ of visits from abroad
- ☐ Someone can answer inquiries in the target language
- ☐ Our products can legally be sold in the target country
- ☐ We'll budget monthly for translation updates
- ☐ We have spec sheets ready to translate
- ☐ We attend overseas trade shows or visit buyers
- ☐ We can update site content ourselves
- ☐ We can wait 3–6 months for results
- ☐ We know which 1–2 languages come first
FAQ
Bottom line: most owners get stuck on how many languages and translation quality; here are the three most common questions.
Should a multilingual site use subdomains or subdirectories?
Most SMEs should use subdirectories (/en/). Authority stays on one domain and maintenance is simpler.
Can we rely on AI translation alone?
For drafts yes, but have native speakers review before launch. Mistranslated specs, units or certifications directly hurt buyer trust.
How many languages should we start with?
Start with 1–2. Pick markets that already send inquiries or trade-show leads, then review after 90 days.
Next step
Bottom line: bring your GA4 country report and last year's inquiries, and we'll help you decide which language comes first. ScriptWalker (a Taiwan-based Laravel/Flutter custom development studio) offers a Multilingual Website Build service: adding one language to an existing site starts at about NT$60,000, and a new 3-language site starts at NT$120,000, including hreflang, sitemaps and inquiry routing. Book a free 30-minute consultation:
- Email: [email protected]
- Phone: 0916-224-047
- LINE: @ufv9089p