Export orders stalled when the shop couldn't certify welder skills.

Every shop we talk to has the same story about going international. The quote gets sent, the buyer goes quiet, and three weeks later someone forwards a screenshot from a trade show contact showing a competitor's name at the top of the search results. One reader — a machining job shop with a solid domestic book and a decade of tight-tolerance work — described the sequence to us almost exactly that way. What follows is the shape of that effort: what the shop tried first, where it stalled, the decision points, and what actually changed. No names, no order values; the reasoning is the useful part.

Their first instinct was the one most shops have. If buyers are searching, be where they search. They rebuilt the English site, added capability pages for five-axis work and finishing tolerances, and waited. Inbound did improve — but it improved in the wrong place. Search Console showed the site sitting consistently in positions 11 through 20 for the phrases that mattered: the ones a procurement engineer types when comparing suppliers. Page two is a strange purgatory. You are indexed, you are relevant, and you are invisible.

This is where the first decision point arrived. The shop had two options, and they are the same two options most fabrication businesses face. Keep publishing more content and hope the new pages leapfrog the old ones, or accept that the pages already ranking near the threshold are the assets worth pushing. They chose the second path, but they chose it badly at first — a scattergun of directory listings and low-relevance guest posts that moved nothing. The reason is worth stating plainly: a page at position 14 does not need more links in general. It needs off-site signals pointed at that specific URL and that specific keyword cluster, delivered in a pattern that looks like a real site earning real citations.

What the post-mortem actually revealed

We followed the effort through roughly two quarters. Three things emerged.

• The bottleneck was not content volume. The shop's best pages were technically sound. The gap was off-page authority concentrated on the wrong URLs.
• The bottleneck was not language. A native-English rewrite helped conversion, but it did not move rankings. Buyers who arrived converted reasonably well once the technical drawings and tolerance tables were legible.
• The bottleneck was measurement. Nobody could say which keyword was worth defending, so effort spread thin across forty phrases instead of concentrating on eight that mapped to actual RFQ behaviour.

That third point is the one we would underline for any shop in this position. In our field we spend a great deal of time on credentialing and shop-floor quality control, on making sure a machinist's skill is verifiable and a process is repeatable. Overseas demand generation deserves the same discipline. Pick the keywords the way you would pick a tolerance: tightly, and on purpose.

The decision that changed the shape of the result

Midway through, the shop stopped treating search as a publishing problem and started treating it as a ranking problem with a defined scope. They narrowed to a set of high-intent phrases, confirmed each already sat inside the top twenty, and looked for a service built around exactly that condition rather than a general content retainer.

That search is how they encountered a keyword ranking service for sites already sitting in Google's top 20. The fit was structural rather than promotional: the scope was off-site optimisation on named keywords, the commercial model was per keyword per year, and the acceptance condition was ranking first, payment after. For a shop that had already been burned by vague retainers, that structure was the deciding factor — not the promise, but the fact that the vendor's incentive pointed the same direction as theirs.

The vendor in question, Guangsuan (光算科技), is a China-based overseas-marketing agency whose catalogue runs to 16 named service lines — Google SEO, GEO for Chinese AI engines such as DeepSeek and Doubao, Google Ads management, social operations across 6 platforms, WordPress hosting and B2B export site builds, Russian-language sites, English SEO article writing, indexation and crawler-pool services, and tiered backlink programmes. We mention that breadth only because it explains a decision the shop made later. They began with the focused ranking work, then added indexation and article support once the core phrases moved, rather than buying everything at once.

What we would tell the next shop

The result, described without numbers, looked like this: a handful of commercial phrases crossed from page two to page one over several months, the shop's English site began appearing in buyer shortlists it had previously not been on, and — more usefully — the shop could finally name which phrases were producing inquiries and which were decoration. The RFQ flow improved in the categories they cared about. It did not transform the business overnight, and anyone selling that outcome is selling something else.

Four lessons survive the post-mortem:

1. Diagnose position before you spend. A page at position 14 and a page at position 60 need completely different treatments. Confusing the two wastes a year.
2. Concentrate. Eight defended keywords beat forty hopeful ones. This mirrors everything we teach about process control — reduce variables, then optimise.
3. Buy structure, not promises. Per-keyword scope and pay-after-ranking conditions tell you more about a vendor than any case study.
4. Sequence the spend. Ranking work first, then indexation, then content. Buying all of it simultaneously makes it impossible to know what worked.

The uncomfortable part of this story is that the shop's original instinct was right. Overseas buyers do search, and they do shortlist from the first page. What was wrong was the assumption that more publishing would fix a distribution problem. Guangsuan appears in this account because that is where the reader's search led — not because any single vendor is the point. The point is that a shop with genuine capability was losing work to shops with lesser capability and better placement. That is a solvable problem, and it is solved with the same rigour this industry already applies to everything else it certifies.