SEO research articles and diagnostic notes for export websites
This page is the index for the research articles on Tianwen SEO Research. It is not a case-study showcase. Every article here is built on something we measured ourselves: Google Search Console exports, indexing tests, AI Overview coverage counts, or a fault we hit on one of the 30–40 export websites we maintain for clients. If a number appears in an article, the article says where it came from and how you can re-run the check on your own site.
The audience is B2B exporters and the people who run their websites: a factory selling custom metal parts, a packaging machine builder, a supplier of waterproof connectors or industrial valves. The questions are always practical — the site is live, so why are there no enquiries; the pages are indexed, so why is the traffic flat; the agency says SEO is done, so how do you verify that without a technical background.
What is published so far in English
The English section is young. Two studies and one method write-up are online; the rest of the research is currently in Chinese and Traditional Chinese, and English versions are added as each study is re-checked against fresh data rather than machine-translated.
- We Tested 500 Chinese Queries: 80.9% Already Return AI Overviews — a coverage count run on 500 Chinese-language B2B queries, with the query list and the counting method included so the result can be reproduced.
- Non-Latin URLs and Google Indexing: An Accidental A/B Test (63 Indexed vs 0) — what happened when the same site published English articles under Latin slugs and under Chinese-character slugs at the same time.
- SEO Audit for a B2B Export Website: The 12 Checks We Actually Run — the checklist we apply to every export site we take over, with the pass condition and the Search Console report for each check.
How to read the research
| Article type | Who it is for | What it settles |
|---|---|---|
| Diagnostic notes | Companies with a live export site and unclear results | Whether the problem is structure, content, a technical fault, or the enquiry path itself |
| Method write-ups | Teams that want to understand how Google finds and ranks B2B pages | How keywords, pages, and enquiries connect — with the data fields to check |
| AI visibility tests | Anyone asking whether ChatGPT, Gemini or Google AI Overviews will mention them | What the AI answers actually cite today, measured in a clean environment |
| Enquiry-path reviews | Sites that get visits but no messages | Where a form, an email route, or a WhatsApp link silently breaks |
Three rules every article follows
The data is ours or it is linked. A claim is either measured on a site we run, or it points to the Google documentation or the third-party page it came from. No industry averages without a source.
Corrections stay on the page. When a later test contradicts an earlier article, the earlier article is updated and the change is noted in the text, not quietly rewritten. In September 2026 two of our own English articles were found to contain factual errors during an external review; both corrections are visible in those articles.
No ranking or enquiry promises. The research describes what we observed. It does not claim that following the same steps will produce the same ranking, traffic, or number of enquiries on another site.
Questions this section is meant to answer
These are the questions export companies actually bring to us, in their own words, and the research is organised around them:
- Our B2B export site has been live for over a year. Pages load fine but we get zero enquiries — how do we find out where the chain breaks?
- My agency says the SEO work is done. I don’t have a technical background — what can I check myself in Search Console in ten minutes?
- We export machinery parts and want to know whether AI search tools will ever recommend a supplier like us, or only the large platforms.
If a question is not covered yet, it goes on the research list. The Chinese and Traditional Chinese indexes of this page list the full set of studies published to date.
