Background
Export-facing trade sites usually fail in two ways. First, the English version is switched by JavaScript on the same URL, so an AI engine only ever crawls one language. Second, the company's identity facts are scattered across the page, so an engine cannot establish who this company is, what it sells and how to reach it — and simply leaves it out of the answer.
What we did
- Built the bilingual site with Chinese on the root path and English on its own
/en/URLs, so both languages are crawlable and indexable - Added hreflang (zh-CN / en / x-default) and self-referencing canonicals so the two language versions declare each other
- Consolidated the identity facts an engine needs — legal name, registered location, contact, scope of business — into structured data
- Rewrote the questions overseas buyers actually ask as answer blocks that can be quoted whole, instead of leaving them inside marketing prose
- Allowed the major AI crawlers and published an llms.txt entry point
Technical approach
| Item | Approach |
|---|---|
| Language routing | Chinese on root, English on separate /en/ URLs (not a JS toggle) |
| Internationalization | hreflang zh-CN / en / x-default, self-referencing canonical |
| Structured data | Organization + WebSite |
| Citable content | Buyer-question answer blocks, each carrying its own subject and qualifiers |
| AI crawling | robots allows the major AI crawlers, plus llms.txt |
Results
| Item | Status |
|---|---|
| Live site | https://www.scimpexp.com/ |
| Languages | Chinese / English on separate URLs |
| Change in AI citations | Not disclosed |
| Enquiry and business data | Not disclosed |
Authorization
The site is publicly accessible; the client name and business data are not disclosed.