PSG Steel Group (Pars Steel Global)

Orbital Web Studio rebuilt PSG Steel Group's WordPress site on Cloudflare Workers, preserving all 81 indexed URLs — including 21 percent-encoded Cyrillic paths — while adding a multi-line RFQ engine to replace a bare phone number.

A site with no headings at all

PSG Steel Group sources and exports Iranian steel and raw materials — 17 product lines, from rebar and wire rod to direct-reduced iron — to buyers across the Gulf, Turkey, South Asia, Africa and the CIS. The old site looked the part, but underneath it was a single Revolution Slider stack: 359 DOM nodes, and not one real <h1> through <h5> anywhere on the site. No document outline means no structure for a search engine, an AI assistant, or a screen reader to read — on a site that otherwise had genuinely good specification content underneath it.

Every one of 81 indexed URLs, kept byte-exact

Most redesigns quietly break a handful of URLs and call it acceptable churn. We treated PSG's 81 already-indexed pages as a binding contract instead: every English and Russian page, decoded and re-encoded, had to match the exact address already sitting in Google's index — including three WordPress transliteration quirks in the Russian slugs that looked like typos but were themselves indexed, so they had to be kept, not "corrected." We verified this page by page across the entire site, not as a spot check.

That lowercase-hex detail mattered more than it sounds: browsers emit uppercase hex by default, but WordPress — and therefore every backlink and every Search Console row already pointing at PSG — used lowercase. Matching it exactly meant zero redirect hops and reporting that lined up from day one, instead of weeks of Search Console noise while Google re-crawled a chain of redirects.

The bug that would have quietly doubled the site

React Router's optional locale segment (/:lang?) is a convenient way to serve one route tree in two languages — until the Russian paths don't share any structure with their English twins, which PSG's didn't. Left alone, that combination would have let /ru/steel-products/rebar/ and the plain English path both resolve as valid — 42 silent duplicate URLs, invisible until Search Console started flagging "duplicate, Google chose different canonical." We caught and closed it before launch with an explicit path-to-page resolver, rather than after the duplicates were already indexed.

Russian rebuilt as a market, not a translation

The old Russian pages used slugs and vocabulary a Russian steel trader would not actually search for — believable evidence of machine translation, not a native buyer's language. Every Russian page on the new site was written from scratch against real trade terminology, not translated from the English draft, the same discipline this case study's own Farsi version follows.

A phone number is not a conversion unit

The old site's only way to start a transaction was a phone number, an email address and a WhatsApp link. For a business where a real enquiry looks like product, grade, standard, tonnage, destination port, Incoterm, payment terms — that is a lot of information to lose in a phone call. We built a proper multi-line RFQ form instead, feeding a real trading-desk pipeline board rather than an inbox.

Six digital hands on the desk

The trading desk doesn't run alone anymore. Six purpose-built agents watch the site and the pipeline around the clock — quoting and acknowledging new enquiries, pulling dated market benchmarks, estimating delivered freight, screening counterparties, drafting technical content from verified specs, and watching all 81 contracted URLs for anything that breaks. Tap any agent below to see what it actually does — every one of them is running in production, not a mockup.

Want this outcome for your business?

If your export or manufacturing business is stuck between "our site was built years ago and it shows" and "we can't afford to lose what we already rank for," talk to us about B2B lead and pipeline systems or get in touch.