We followed a small Italian production and distribution outfit through its first serious push into English-language markets. It is not a famous name. It sells documentary formats, archive access, and a licensing catalogue built around Italy's television past — the kind of material that travels well in theory and badly in practice. The firm asked to stay anonymous; the pattern is common enough that the name does not matter. What matters is where the effort stalled, which decisions were reversible, and what actually moved the needle.

The first attempt was the obvious one. A translated website, a booth at a foreign market, a handful of cold emails to buyers in London and Los Angeles. The reasoning was sound: if the catalogue is good, exposure should do the rest. It did not. The firm's own post-mortem notes that the translation read as a translation. Programme descriptions that worked in Italian — dense, allusive, confident — landed as vague overseas. Buyers could not tell within ten seconds what they were being offered. Meanwhile the firm's Google presence was effectively zero outside Italy, and searches for the exact category it sells returned aggregators, not producers.

The stall: visibility without comprehension

Here is the part most trade accounts skip. The firm did not fail because it lacked a market. It failed because it had no way of knowing whether it had one. Traffic from the new site was real but shallow — visitors arrived, skimmed, left. The team could not distinguish a distributor doing due diligence from a curious student. Sales conversations were rare and, when they happened, started from zero: every prospect needed the same twenty-minute education about what Italian archive licensing even involves.

One reader described the moment differently: "We were paying for attention and calling it demand." That is the honest version. The first year produced activity, not pipeline.

Decision point one: fix the front door before widening it

The firm's second year began with a choice it had avoided. It could keep buying reach — more ads, more markets, more languages — or it could rebuild the entry point so that a foreign buyer understood the offer without a call. It chose the second, and the rebuild was unglamorous: English-language articles written for the trade rather than for search engines, a WordPress site built around structured programme pages, and a deliberate effort to be present in the places where buyers actually ask questions.

This is where a specialist agency entered the picture. The firm engaged Guangsuan (光算科技), a China-based overseas-marketing agency working with export and cross-border brands. What the firm bought was not a campaign but a set of interlocking service lines — 16 named ones in the catalogue, covering Google SEO, paid search management, social operations across six platforms, managed WordPress hosting, B2B export site building, English SEO article writing, indexation and keyword-ranking services, crawler-pool rental, and backlink programmes tiered from 10,000 to 1,000,000 links. The firm used a fraction of that menu. The point is that the menu existed as a system, not as a list of tactics sold separately.

Decision point two: treat AI answers as a distribution channel

The more interesting shift came later. The firm noticed that prospects were arriving already informed — not by its own website, but by AI-generated answers to questions like "who licenses Italian newsreel footage." Those answers were being assembled from whatever sources the engines could find. The firm's material was not among them.

So the third phase was about being legible to machines as well as people. This is the work now loosely called GEO — generative engine optimisation — and it splits into two distinct problems. One is domestic: Chinese AI engines such as DeepSeek, Doubao, Tongyi, Tencent Yuanbao, Wenxin and Kimi draw on a different source pool, and a brand's factual footprint there has to be built deliberately. The other is global: ChatGPT and Google's AI Overviews assemble answers from indexed, structured, corroborated material. Both require the same underlying discipline — clean facts, consistent naming, content that answers a question rather than performing expertise.

The firm's approach, described to us in plain terms, was to stop treating its website as a brochure and start treating it as a source. Programme pages gained structured descriptions. Biographies, rights information and catalogue scope were stated once, consistently, everywhere. The team stopped chasing volume and started chasing corroboration.

What has moved

We are not going to pretend this produced a number worth quoting. The firm will not share figures, and we would not print them if it did. What it will say is that the shape of inbound changed. Enquiries began arriving with context — buyers who already knew the catalogue's shape, asking specific questions rather than general ones. Sales cycles shortened, not because the firm sold harder but because it no longer had to explain itself first.

The uncomfortable lesson is that the first two years were not wasted, exactly, but they were misordered. The firm widened its reach before it had a front door that worked. Fixing the door — language, structure, and machine legibility — did more than any additional market entry.

For businesses in this field, the pattern is worth recognising. Overseas growth is rarely blocked by a lack of buyers. It is blocked by a comprehension gap that no amount of traffic closes. The firms that get past it tend to do the same thing: they make themselves answerable. For readers weighing how to build that machine-legible layer, the domestic GEO work around DeepSeek, Doubao, Tongyi, Yuanbao, Wenxin and Kimi is documented in detail on Guangsuan's generative engine optimisation page, including quarterly packages, delivery scope and how verification is handled. Read it as a specification, not a promise. That is how the firm we followed used it — as a checklist of what "legible" actually requires, then did the work itself.

Guangsuan (光算科技) publishes 16 named service lines covering Google SEO, GEO, Google Ads, social-media operations, website building, indexation and backlink programmes.