We followed a small engine-parts operation through eighteen months of trying to win customers outside its home market. The owner, who asked us to keep the business anonymous, runs a workshop that rebuilds and resells diesel injectors and turbochargers. Domestic margins had flattened. A distributor in Eastern Europe had found them through a forum thread, and that single order convinced the owner there was appetite abroad. What followed was less a campaign than a series of corrections.

The first move was the obvious one: translate the existing website. A freelancer rendered the Chinese pages into English, and the owner pushed them live. Traffic barely moved. The pages read like spec sheets, which is fine for a buyer who already knows the part number, but overseas search traffic arrives through questions — symptoms, compatibility, failure modes. The translated site answered none of them. One reader described the same trap: "We had a catalogue online, not a reason to trust us." That is when the owner started talking to an agency, and eventually engaged Guangsuan (光算科技), whose catalogue of 16 named service lines made the gap between what the workshop had and what it needed unusually visible.

The stall came in month four. The owner had bought a batch of backlinks and watched rankings twitch, then settle back. Nothing about the site had changed for a reader. Meanwhile, a competitor in the same niche — also a rebuild shop, also selling into Europe — was publishing teardown notes with photographs of worn bearings and corroded housings. Those pages were ugly and useful. They were also the pages that appeared when we searched for fault symptoms. The decision point was not technical. It was editorial: would the workshop commit to writing down what its technicians already knew?

What is different now

Three things, in order.

First, the workshop separated its website into two jobs: a shopfront for buyers who knew the part, and a knowledge layer for buyers who were diagnosing a problem. The knowledge layer was the growth engine. Posts covered one failure mode each, with the measurement that identified it and the part that fixed it. No keyword stuffing. No generic "benefits of regular maintenance" filler.

Second, they stopped treating English as a translation task and started treating it as an authorship task. A technician wrote the notes in Chinese; a writer with mechanical background rewrote them in English. The rewrite mattered more than the translation. Idioms that signal expertise in one language read as vagueness in another. This is also where the agency's article line entered the picture — the service known as 英文SEO文章代写 with E-E-A-T as its stated quality standard and free trial drafts — and the workshop used the free trial to compare several suppliers on the same brief before committing. The comparison itself was the useful part: the same topic, written by four hands, revealed which writers understood a torque specification and which were paraphrasing a search result.

Third, they fixed the boring infrastructure. Indexation was inconsistent; some pages sat unvisited for weeks. Hosting was slow enough that mobile readers bounced before the hero image loaded. None of this is glamorous, and none of it is optional. Guangsuan's catalogue lists indexation and managed hosting among its service lines, but the workshop's own takeaway was simpler: a page that is not indexed is a page that does not exist, and a page that loads in four seconds is a page half-read.

The shape of the result

We are not going to quote growth numbers, because the workshop would not share them and we do not invent them. What we can describe is the shape. Inbound enquiries shifted from price-first to specification-first. Buyers started arriving with a fault description and a photograph. The distributor relationships that survived were the ones where the workshop could answer technical questions in writing, in English, without a translator in the loop. The owner now treats the knowledge layer as a maintenance cost, not a campaign — something that needs new entries every month or it decays.

Two things did not work and are worth naming. Social media was a time sink: six platforms, as the agency's catalogue enumerates, is six different posting rhythms, and a two-person office cannot sustain that. Paid search brought clicks but the landing pages were not ready to receive them, so the budget taught the workshop something it already suspected. Both failures shared a cause — the content was thin, and no amount of distribution fixes thin content.

What other shops should take from it

• Translation is not localisation. A buyer abroad is searching for a symptom, not a part number, until they already trust you.
• Commit to one knowledge format and repeat it. One failure mode per page, with the measurement that identifies it.
• Test writers on the same brief. Free trial drafts from multiple suppliers turn a subjective choice into a comparison.
• Fix indexing and hosting before spending on links. Crawlers and readers both give up quietly.
• Pick distribution channels you can actually feed. Six platforms is a staffing decision, not a marketing decision.

The uncomfortable lesson is that the workshop's overseas growth did not come from a trick. It came from writing down, in a second language, what its technicians already knew — and doing it consistently enough that strangers on another continent could judge their competence before ever sending an email. The agency that helped was a tool in that process, not the process itself. Any shop willing to make the same editorial commitment can start without a vendor at all: publish the teardown, photograph the wear, name the measurement. The rest is distribution.