Case Study: Efecan Polisaj

Context

An export business that needed a website to match

Efecan Polisaj has been making polishing buffs for jewelers and metalworkers out of Topkapı, Istanbul since 2002. Three generations of the family have run the business. They export to twenty-five-plus countries.

What they didn't have was a website that reflected that. The previous one was a basic page that didn't really speak to a buyer researching them from Germany, Brazil, or the Gulf. The brief was simple to describe and harder to execute: a B2B catalog that works in nine languages, looks professional enough for German jewelry tool wholesalers and global metalwork distributors to take seriously, and shows up when a buyer in São Paulo searches "polishing buff manufacturer Turkey."

No checkout. B2B sales happen over WhatsApp anyway. Just a proper shopfront that performs in nine markets.

What I Built

Multi-language Structure

Right-to-left layout for Arabic, with the whole interface flipping (icons, alignment, text direction)
Per-language fonts that pair visually instead of clashing
Per-market preview cards so links shared into a German Slack don't show Turkish copy
Language tags and sitemaps that tell Google which page to show which audience

Industry Terminology Research

Original "Reinforced Polishing Buff" was technically correct, commercially nonsense
Industry trade term: knife edge buff
Same exercise across all 9 languages, mapped to local trade vocabulary
"Extra Quality" rendered as Premium in RU/AR, kept Latin in DE/ES, per local convention

Photography Workflow

Custom pipeline for 100+ workshop product shots
Background removal so photos drop cleanly onto any card colour
Programmatic homepage hero composition
Brushes category solved by re-designing the card around the photo

Three Hand-crafted SEO Passes

Turkish: toptan cila bezi, kuyumcu cila bezi, cila bezi imalatçısı
English export: manufacturer-intent searches, Turkey origin signal
German (DACH): added missing Schwabbelscheibe, Großhandel, Türkei, Crown
Origin signal rotated across 32 product descriptions, varied phrasing

AI-assisted SEO

Italian, Spanish, French (rest of Europe)
Russian (CIS demand)
Arabic (Gulf and North Africa demand)
Generative AI tools paired with terminology research, validated against actual market vocabulary
Per-language meta titles, descriptions, product-level keywords

Brand System

Navy #14233F, brass #C9A15A, cream #FBF8F1
Fraunces display, Manrope body, JetBrains Mono mono
Brass corner ticks framing every product image
Content updates reviewed before going live

Industry Terminology

From "Reinforced" to "Knife Edge"

This is where most multi-language sites fall over. The original translator had rendered "destekli cila bezi" as "Reinforced Polishing Buff." Technically correct. Commercially nonsense. The term used in the polishing trade is knife edge buff. A German jewelry wholesaler reading "Reinforced Polishing Buff" sees an importer; reading "Knife Edge Buff" sees a manufacturer. Every language got the same treatment based on terminology research.

Locale	Translator's first pass	Industry term
EN	Reinforced Polishing Buff	Knife Edge Polishing Buff
DE	Verstärkte Polierscheibe	Spitzkanten-Polierscheibe
FR	Disque de polissage renforcé	Disque à polir tranchant
IT	Mola lucidante rinforzata	Disco lucidante a tagliente
ES	Disco pulidor reforzado	Disco pulidor de filo
PT-BR	Disco de polimento reforçado	Disco polidor de aresta viva
RU	Усиленный полировальный круг	Полировальный круг с заострённой кромкой
AR	قرص تلميع مُدعَّم	قرص تلميع بحافة مدببة

Same exercise for "Extra Quality." Translated semantically where the language conventions called for it (Premium in Russian and Arabic, "tranchant" in French), kept in Latin script where it reads as a brand-tier descriptor (DE, ES). Per-language decisions, not search-and-replace.

Three Hand-crafted SEO Passes

One pass would have left two markets cold

The English-speaking export buyer searches differently from the German wholesaler. The Turkish reseller searches differently from both. So three separate audits, three different keyword lists, three different metadata strategies.

Turkish. Optimised for the wholesale-intent terms a Turkish buyer would type: toptan cila bezi (wholesale polishing buff), kuyumcu cila bezi (jeweler's polishing buff), cila bezi imalatçısı (polishing buff manufacturer). Cleaned up old references to a discontinued product line and pointed everything at the current catalog.

English. Targeted manufacturer-intent searches: polishing buff manufacturer, wholesale polishing buffs, Crown Extra Quality (the company's own line). Meta descriptions explicitly call out Turkey as origin, since DACH and US import buyers commonly search by manufacturing country.

German. The most surgical pass. The German market uses two equally common synonyms in the polishing and jewelry trade: Polierscheibe and Schwabbelscheibe. The site only had one of them. Same gap with high-intent keywords like Großhandel (wholesale), Türkei (origin), and Crown (the company's own line). I rewrote German metadata across five page templates that had been falling through to a generic title, added the missing keywords without stuffing, and made sure each product description carried an origin signal. Each German buyer landing on any product page now sees a consistent "manufacturer in Turkey" signal, varied across products so it doesn't read templated.

Großhandel Polierscheiben aus der Türkei seit 2002. Crown Extra Quality, farbige erstklassige Schwabbelscheiben in 7 Härtegraden, importierte Polierpasten von Menzerna und Dialux. Direkt vom Hersteller.
Sample DE meta description after the pass

That's a marketer translating "we want to rank for German buyers" into actual on-page copy that German buyers will type and Google will reward.

AI-assisted SEO

Italian, Spanish, French, Russian, Arabic

The remaining markets got an AI-assisted pass rather than a hand-crafted one. Italian, Spanish, and French covered the rest of Europe. Russian and Arabic covered the CIS and Gulf / North Africa demand. Each language ran through generative AI tools combined with terminology research: feed in industry context, generate candidate trade terms, validate against actual market vocabulary and search behaviour, then map the on-page copy to match.

Each language got per-page meta titles, descriptions, and product-level keywords aligned to local trade language. A buyer in Milan, Madrid, Lyon, Moscow, or Riyadh now lands on copy that reads like trade vocabulary, not translated marketing-speak. Same standard as the manual passes, faster cycle, and the AI worked because the terminology research was real, not invented.

Outcomes

What shipped

Nine languages, with Arabic in proper right-to-left layout. Sister sites in this segment usually run two or three languages and break in Arabic.
Around 370 catalog pages, served fast from a global network. No perceptible speed difference between São Paulo and Munich.
Three hand-crafted SEO strategies (Turkey, English export, DACH) plus AI-assisted SEO across Italian, Spanish, French, Russian, and Arabic, all built on real terminology research.
Industry-correct terminology in every language, copy that wouldn't make a German wholesaler smirk.
Custom photography workflow that turned shop-floor JPGs into a coherent product catalog.
Content updates go through review before going live, no accidental publishes.

Key Takeaways

What this project confirmed

Industry terminology is the difference between sounding like an importer and sounding like the manufacturer. Knife edge buff vs reinforced polishing buff isn't a small detail. It's who the buyer thinks they're dealing with.
SEO is per-market, not per-site. Turkey, anglophone export buyers, and DACH each got their own keyword list, metadata, and origin signals. One pass would have left two markets cold.
Multi-language done properly is a market-research project, not a translation project. Picking the right term in eight languages took longer than building the site.
AI is a serious efficiency lever when paired with proper research. Used generative AI for terminology and SEO copy across five additional languages, after the industry research was solid and the validation criteria were clear. The strategy was already there; AI accelerated the output.
Brand and substance go together for B2B. The catalog had to look as serious as the company is. Workshop photography, brand colour discipline, and product page hierarchy all do that work.