SEO Audit for a B2B Export Website: The 12 Checks We Actually Run

An SEO audit for a B2B export website is not a 90-page PDF with a score out of 100. It is twelve checks, each of which either passes or fails on evidence you can open yourself in Google Search Console. This page lists the twelve we run on every export site we take over, what “fail” looks like in each one, and which of them actually decide whether a manufacturer gets enquiries. The list comes from maintaining 30–40 client sites and from the numbers on our own three sites, not from a tool’s default report.

Why most SEO audits for exporters are useless

The typical audit is generated by a crawler and lists 400 “issues”: missing alt text on 212 images, 37 title tags over 60 characters, 9 pages with duplicate meta descriptions. All of that can be true and the site can still be invisible, because the crawler cannot see the three things that matter for an export site: whether Google is indexing the product pages at all, whether the pages that do rank are the ones a buyer would land on, and whether an enquiry submitted from the site reaches anyone.

A concrete example from our own property. Over 16 months, tianwenwangluo.com received 543,305 impressions and 7,460 clicks in Google Search — an average CTR of 1.37% at an average position of 15.8. A crawler audit of that site scores well. The Search Console data says something different: most of those impressions were on pages that have nothing to do with the service we sell. The audit that matters starts from that table, not from the crawl.

The 12 checks, in the order we run them

1. Is the site indexed at all, and which URLs?

Open Search Console → Pages (Indexing). Read the “Not indexed” reasons, not the total. For an export site with 200 product pages, the failure pattern is usually “Crawled – currently not indexed” on exactly the product pages, while the blog is fully indexed. Then spot-check five product URLs with the URL Inspection tool; it shows the canonical Google chose, which is frequently not the one you expect.

2. Are the indexed URLs in the right form?

Non-Latin slugs on English pages are a silent killer. We measured it by accident: on one of our sites, English articles published with Latin slugs were indexed 63 times; the same type of article published under Chinese-character slugs was indexed zero times over the same period. The full write-up is at Non-Latin URLs and Google Indexing: 63 vs 0. Check every English URL on the site for percent-encoded characters.

3. Does one page own each buying query?

Export sites built from a keyword list end up with three pages for “custom metal parts”: a product page, a category page and a blog post. In Search Console → Performance, filter one query and look at the Pages tab. If two or more URLs alternate for the same query week to week, Google is choosing for you, and it usually picks the weaker page.

4. Does the query table match the product list?

Export the top 500 queries and mark each one as buyer intent, job-seeker intent, student intent or noise. On many industrial sites, fewer than 10% of impressions are from queries a purchasing manager would type. That single number tells you whether content or technology is the problem.

5. Is the language and region markup consistent?

If the site has Chinese, Traditional Chinese and English versions, every page needs a self-referencing canonical and reciprocal hreflang tags. Google’s requirements are specific: the tags must point both ways, and each URL must return 200. See Google’s localized versions documentation. We found and fixed this exact fault on our own research index this week — the English URL carried a canonical pointing at the Chinese page, so Google had no English version to rank.

6. Core Web Vitals on a product page, on mobile

Not the homepage. Google’s thresholds are LCP at or under 2.5 seconds, INP at or under 200 milliseconds and CLS at or under 0.1, measured on real users where data exists (web.dev: Web Vitals). A 15 MB hero image on the product template fails all three for every product page at once; that was the actual bottleneck on one client’s site, not the server.

7. Structured data that describes the business, not just the blog post

Organization markup with legal name, founding date and external identifiers is what lets Google — and AI answer engines — connect the site to a real company. Article schema on blog posts does nothing for a supplier’s credibility. Check the homepage JSON-LD for exactly one Organization node.

8. The enquiry path, end to end

Submit the contact form from a phone. Check the inbox and the spam folder. Click the WhatsApp link. On a site we audited in September 2026, the form had been silently failing for 12 days because the reCAPTCHA site key was registered to a different domain — 207 paid clicks landed on a form that could not send. No SEO metric shows this; only the test does.

9. What Google’s own report says about AI Overviews

For Chinese-language queries the exposure is already large: we tested 500 B2B queries and 80.9% returned an AI Overview (method and query list). Check whether the site’s key pages are being cited in those answers or displaced by them; the Performance report’s search appearance filter is the starting point.

10. Internal links from pages that already rank

List the ten pages with the most impressions. Count how many of them link to a product or service page. On most export sites the answer is zero — the traffic lands on a tutorial and leaves.

11. Crawl waste

Faceted navigation, tag archives, paginated comment feeds and search result pages can consume more crawl than the product catalogue. Compare the number of URLs in the sitemap against the number Google reports as discovered. A 5:1 ratio of discovered to submitted usually means the catalogue is being starved.

12. Who the site says it is

Founding year, company name, address and the person behind the site should be identical across the website, the business registry and any third-party listing. An AI assistant checking a supplier will read all three; when they disagree it says so. This is the cheapest check on the list and the one most sites fail.

What a pass looks like

How long it takes and what it costs

The twelve checks take one working day for a site under 500 URLs when Search Console access is granted. Without Search Console access, checks 1, 3, 4, 9 and 11 cannot be done honestly; a report that claims to do them from the outside is guessing. We do not publish a price for the audit on this page; the scope depends on the number of languages and product pages, and the quote is written per site. What we can say is that the audit is the entry point, not the service — the value is in the fixes, and most sites need fewer than five of them.

If you want to see the method before hiring anyone, the studies behind these checks are on the research index, and the service description is on tianwenwangluo.com.