Lighthouse is hiring sellers who work in Czech, Hungarian, Dutch, Bahasa, Japanese and Korean, and an AI engineer to automate marketing. Its site already has seven locales, but only the product pages are translated: in a sample of 24 blog posts and customer stories, the German and Japanese copies were the English text, each marked as a German or Japanese page with its own canonical. That is about 7,000 English pages that tell search engines and AI assistants they are local-language content, including a Japan cherry blossom demand report on /ja/ that a Japanese hotelier cannot read. Lighthouse sells Connect AI on the promise that AI platforms reward "accurate, well-structured content"; its own localised resources fail that test.
First action
Export the resource pages per locale and flag which are identical to English.
How to measure it
Local-language organic sessions to /resources
- Now
- –
- Target
- +50%
- By
- 2 quarters
Steps
Measure the gap
Compare every locale page with its English original; list identical copies.
Marketing Operations Manager 2 days
Stop the duplicates
Canonical untranslated copies to English and remove them from the locale sitemaps.
Web Lead 1 week
Build the workflow
Agent translates from the CMS with a glossary of hotel revenue terms.
Senior AI Solutions Engineer for Marketing 2 weeks
Set review
Native-speaking sellers approve each page before it goes live.
Customer Success Lead ongoing
Translate what sells
Customer stories, independent-hotel guides and local market reports first.
Content Marketing Lead 1 month
Use in outreach
Local sellers link the translated pages in sequences.
Sales Development Representative - Dutch Speaker ongoing
Measure
Indexed pages, local organic sessions and demo requests by locale.
Marketing Operations Manager monthly
Why it works
Make this the first project for the new AI Solutions Engineer for Marketing. Build an agent workflow that translates resource pages from the CMS into each locale, with a hospitality glossary and a review step by a native-speaking seller. Start with the pages that sell: the 105 customer stories, the independent-hotel guides and market reports for the locale's own country. Until a page is translated, point its local copy's canonical to the English page and drop it from the sitemap, so only real translations are indexed.
Local-language search and AI answers become a lead source for each new seller, and the market reports and customer stories they send in outreach are finally in the buyer's language. Removing thousands of duplicate English pages also concentrates ranking signals on the originals.
Ready to use
ChecklistLocale clean-up list
[ ] List locale pages identical to English [ ] Canonical untranslated copies to the English page [ ] Remove untranslated copies from locale sitemaps [ ] Keep hreflang only between real translations [ ] Translate meta descriptions on product pages (e.g. /ja/platform/independent-hotels) [ ] Customer stories first, then guides and market reports [ ] Native-speaker review before publishing
Risks and how to handle them
Machine translation reads badly in Japanese
Native review on every page; glossary maintained by the North Asia team.
Short-term traffic dip when duplicates are canonicalised
The English originals keep the signals; track them in Search Console.
Evidence
- Your websiteLocalised resources (checked October 21, 2026): every page has copies under /de/, /es/, /fr/, /it/, /ja/, /nl/ and /pt-br/, each with its own canonical, an html lang tag for that language and hreflang links to the others. In a sample of 24 blog posts and customer stories, 23 of the German copies and all 24 Japanese copies were the English text word for word, including "Japan cherry blossom season: Hotel demand and pricing data" on /ja/. Product pages such as /ja/platform/independent-hotels are translated, though its meta description is still in English.Source
- Your websiteSitemap (8,880 URLs, read October 21, 2026): 1,039 English resource pages (589 blog posts, 105 customer stories, 186 events, 71 insights, 77 podcast episodes, 9 Lighthouse Live, 1 tool), each repeated in 7 locales (1,109 URLs per locale). No Korean, Indonesian, Malay, Czech or Hungarian locale. Customer stories from those markets exist in English only, e.g. La Casa Hotel Seoul, Lotte Hotels & Resorts, Potato Head Suites & Studios, Cititel Hotel Management and AC Hotel Bratislava Old Town. 18 /lp pages include a ChatGPT bookings demo, refer-a-friend and partner referral forms for Mews, SkyTouch and Shiji.Source
- HiringLocal-language sales hiring: an Account Executive - Czech or Hungarian Speaker (Madrid and Barcelona, posted October 9 and 20, 2026), a Sales Development Representative - Dutch Speaker who writes "highly personalised emails in Dutch and English" and will "Work closely with Marketing to generate leads and follow up after trade shows, webinars, events and campaigns", Business Development Managers for Indonesia and Malaysia ("Native Bahasa speaker"), an Area Director North Asia for Japan and Korea ("Fluent in English and Japanese; Korean is a plus", about 80% Japan) and UK Account Executives.Source
- HiringSenior AI Solutions Engineer for Marketing (Ghent and Barcelona, posted October 2, 2026): will "Design, build, and operate AI agent workflows and integrations to optimize our marketing operations for scalability" and "co-create the roadmap for AI adoption"; the marketing team "continually look[s] for points of friction in the funnel that marketing can help to solve" and is "on the lookout for new growth levers". A sister role, Senior AI Solutions Engineer for Revenue & Go-To-Market, builds agent workflows for the sales teams.Source
- Your websiteConnect AI page: "ChatGPT has 900 million weekly users" (the /lp/chatgpt-bookings-demo page says "800 million"); "Hotels with Connect AI see 21% more AI traffic"; GEO: "AI platforms constantly crawl the open web and what they find shapes their answers", so hotels need "accurate, well-structured content"; AI Visibility Insights track mentions in ChatGPT or Gemini and benchmark competitors; placement in The Hotels Network app in ChatGPT.Source