WordPress to Next.js: 15× organic traffic.

The growth happened over six months. The SIM Printing Suppliers engagement ran 26 weeks in total, and the traffic increase came from replatforming and the content programme running together. Replatforming alone would not have produced it: a faster site with nothing new to publish does not move rankings. The two workstreams advanced on one schedule.

The WordPress stack could not carry the load of speed and multiple languages. The site was slow and did not meet the requirements of a five-language export site, with each new language landing as a patch on the existing install. On the new build page load was brought under one second and the threshold held after launch, because performance kept being measured.

Two requirements drove the choice: five languages living in one architecture, and load time under measurable control. The site was rebuilt as a web application, with language structure, content model and performance budget all part of the same decision. The technology was picked last, not first — the export goal and the depth of content were written down before the tool was named.

The languages were not added later; all five were designed at once. No translation layer was bolted onto an existing structure — each language was built from the start with its own content structure. The approach opened the door to demand from outside Türkiye and wrote the fact that local search languages differ into the architecture itself. A language added afterwards usually stays half-finished.

The measure expresses the brand's visibility across AI engines, and it went from zero to 40,000. It is a separate indicator from classic search ranking; the same content is tracked separately on the search engine side and the AI engine side. Starting from zero matters: the brand had never appeared as a source in those engines, because there was no written content to cite.

Content was written with a table of contents, Q&A sections and self-contained passages. In the technical guide format, each section gives a complete answer when read alone, and that passage is the unit an AI engine cites. Technical topics such as printing chemicals were structured to open with the reader's question rather than with the language of a product catalogue.

The knowledge became content by being written down. The accumulation from production running since 1983 existed nowhere in writing, so there was no source for a search engine or an AI engine to cite. The product catalogue, the search language of target markets and the technical topics competitors had left empty were mapped, and the content programme was written straight against that map.

Performance is lost after launch day rather than on it. Page load was brought under one second and the threshold held after launch, and because measurement continues, any regression is corrected as soon as it appears. On a five-language site this is also an export question: an overseas visitor on a slow connection leaves before the page renders.

The structure opened demand from outside Türkiye. In the founder's words, a company doing the same work for over forty years began receiving requests from all over the world for the first time. Building each language around its own search vocabulary was decisive here: the same product is searched under different names in different markets, and translation alone does not close that gap.

The approach works where technical knowledge exists but has never been written down. At SIM the lever was knowledge the company already held; content only made it findable. On a catalogue with no technical depth to explain, the same programme does not deliver at this scale. The second condition is an in-house expert to feed the content — technical text written from outside does not carry that depth.

The work falls under custom software and mobile apps. The five-language architecture, the content programme and the search side ran as parts of one engagement. No off-the-shelf platform was installed; the site was written from scratch as an application, because the goal was not simply a new design but measurable speed together with a multilingual content structure.