The Multilingual Website Guide
“Let’s just add an English version” sounds like an afternoon’s work in most projects. 🌍 It is not: a second language is a small second website — with its own addresses, its own search visibility and its own upkeep.
We do not know this from books; we know it from our own site: we run our pages and content series in two languages, and most warnings in this guide were distilled from our own scars. 🧪
The guide answers four questions: when a second language is genuinely needed, what separates translation from rewriting, which technical decisions are critical, and how a bilingual site stays alive. 🧭
Honesty first: not every site needs a second language. A language added for the wrong reason decays into dead, unupdated pages — and dead pages damage reputation. This is a cost decision and deserves to be made like one. ⚖️
When Is a Second Language Justified? 🎯
BU BÖLÜMÜN ÖZETİ
- Foreign customers are already knocking
- The data shows foreign visitors
- A target market has been chosen
- Filter out the false reasons
The decision rests on one of three pieces of evidence, not on enthusiasm. Without evidence, the second language remains an expensive ornament — and ornaments demand upkeep.
The evidence. 🔎
Foreign customers are already knocking
The strongest evidence is existing demand: emails, calls and messages arriving in another language. 📨 Demand without a language version means customers wrestling with translation tools — some do not wrestle; they leave. Here the second language is not an investment; it is plugging a leak.
The data shows foreign visitors
If analytics show a meaningful share of visitors from abroad, demand is announcing itself. 📊 Check the country breakdown: the language candidate is wherever the density sits. The assumption that “English covers everywhere” does not hold in every market.
A target market has been chosen
The third evidence is strategic: the business has decided to enter a specific foreign market. 🧭 Then the language is part of the marketing plan and gets budgeted with it. No decision, no language; a “maybe someday” language can wait for its proof.
Filter out the false reasons
“The competitor has one”, “it looks professional”, “maybe tourists will browse”: none of the three is evidence. 🚫 You cannot know whether the competitor’s English pages win any business — the same trap we dismantled in the homepage guide. An evidence-free language is the most expensive window decoration there is.
Not Translation — Rewriting ✍️
BU BÖLÜMÜN ÖZETİ
- The reader’s question changes
- Examples and context localise
- Machine translation is a draft, not a publication
- A half language is full damage
The biggest multilingual illusion lives here: having the text translated does not finish the job. A foreign reader asks different questions and carries different worries — the text gets rebuilt around them.
Four lessons from our own experience. 🧪
The reader’s question changes
A local reader can place you on the map; a foreign reader begins with “can I trust this company, and how would working together even look?” 🌐 So the foreign version pulls trust signals, the working process and communication logistics to the front. This is exactly why we did not translate our own English pages — we rewrote them.
Examples and context localise
Local institution names, local payment habits, references to local regulation land on nothing in a foreign reader. 🗺️ Rewriting swaps them for universal equivalents. The practical rule: every proper noun the reader will not know is an explanation debt — pay it or cut it.
Machine translation is a draft, not a publication
Automatic translation produces a fast first draft; it must not go live without an editor. 🤖 Half-correct text is worse than no text: language errors read as workmanship errors. The publishing bar: a final read by someone who owns the language.
A half language is full damage
The worst outcome: a patchwork page with an English menu and half-translated content. 🩹 Visitors spot the patches instantly and leave. The principle is firm: fewer pages, fully done beats many pages half done — keep the scope small, keep the integrity whole.
Technical Setup: The Critical Decisions 🧱
BU BÖLÜMÜN ÖZETİ
- Each language on its own path
- Counterpart pages linked together
- The switcher must lead to the twin
- The template is part of the language
On the technical side, multilingualism stands on a handful of critical decisions. The table lists them; correcting a wrongly built structure later costs as much as a migration.
The decisions. ⚙️
| Decision | Sound practice |
|---|---|
| Address structure | Each language lives on its own address path |
| Language linking | Every page technically linked to its counterpart |
| Language switcher | Visible, and it leads to the matching page |
| Auto-redirects | Suggest, never force; the reader keeps the choice |
| Menus and template | Buttons, menus, footer — every piece in that language |
Each language on its own path
Every language needs a separate, consistent address path; structures that switch language on one address cannot present two languages separately to search engines. 🛣️ This choice is made at the start; changing it later means moving the whole accumulated history.
Counterpart pages linked together
Each page must be technically linked to its equivalent in the other language — that is how search engines learn which page pairs with which. 🔗 Without the link, the two languages can be treated as each other’s competitors. In our own setup we folded this link into the publishing pipeline itself: a new page gets connected to its pair the moment it goes live.
The switcher must lead to the twin
A visitor pressing the language button should land on the counterpart of the page they were reading — not get thrown to the homepage. 🎯 The homepage-throw is the fastest way to lose someone’s place. Where no twin exists, the switcher leads to the nearest parent page.
The template is part of the language
Content gets translated while menus, buttons, form labels and the footer stay behind — the half-finish we encounter most. 🧩 To the visitor, a Turkish menu on an English page is a stamp of carelessness. The pre-launch walk must cover the template pieces too.
Budget and Effort: The Honest Arithmetic 💰
BU BÖLÜMÜN ÖZETİ
- Setup is one-off
- Content is the largest item
- Continuity is the forgotten line
- Start small, finish whole
The cost of a second language never fits one line; it splits into setup, content and continuity. Doing the full arithmetic upfront protects against the half-finished language.
The three items. 🧾
Setup is one-off
Address structure, language pairing, the switcher, template translation: built once and, built right, forgotten. 🏗️ Stinginess here is expensive; correcting a wrong address structure later costs as much as a migration.
Content is the largest item
Rewriting the existing pages takes the biggest share of the budget — the effort multiplies by page count. ✍️ That is why the scope list comes before the money: starting with fewer pages shrinks the budget too.
Continuity is the forgotten line
Every new piece’s second-language step is a permanent extra effort and deserves its own budget line. 🔁 Forget the line and the flow snaps in the first busy month; the patchwork site is that forgotten line’s invoice.
Start small, finish whole
The right strategy is integrity over scale: launch with few pages, keep them complete, expand as evidence arrives. 🌱 The path protects budget and reputation alike; it always beats launching big and patching.
How a Bilingual Site Stays Alive 🌱
BU BÖLÜMÜN ÖZETİ
- Production tied to the flow
- Scope chosen deliberately
- Two languages, two scorecards
- Let’s plan your multilingual structure together
Setup happens once; the real exam is continuity. The visual shows the bilingual publishing flow — a two-language site run with one-language habits inevitably turns to patchwork.
Four steps. 🔁
Production tied to the flow
Bilingualism survives on flow, not willpower: new content does not count as “done” until the second-language step is complete. ✅ We made this a rule in our own content series; article pairs are produced together and published together. Once exceptions accumulate, the patchwork has begun.
Scope chosen deliberately
Not every page needs a twin: which pages live in both languages gets listed at the start. 🗂️ Services and contact get paired; a local campaign page may not. A deliberate gap is not a problem; an unplanned one is.
Two languages, two scorecards
A single combined number hides both languages’ truth: visits, forms and conversions get tracked per language. 📊 If the second language’s form never fills, the problem may sit not in the language but in that version’s trust signals — and without split measurement nobody ever sees it.
Let’s plan your multilingual structure together
If a second language is on your mind, reach us via Web Design; we will test the decision against the three evidences and plan the critical setup steps with you. If the evidence is missing, we will say so — we ran this exact journey on our own site and know the price of an unnecessary language. 🚀
Frequently Asked Questions 💬
Sık Sorulan Sorular
If one of three evidences exists: incoming foreign-language demand, a meaningful foreign visitor share in the data, or a made market-entry decision.
Because the foreign reader asks different questions; the text gets rewritten with trust signals and working process pulled forward.
As a draft, yes; it must not go live without a final read by someone who owns the language.
Each language on its own separate path; single-address language switching cannot present the languages separately.
Technically linking each page to its counterpart; without it, the two languages can compete against each other in search.
To the counterpart of the current page; throwing visitors to the homepage loses their place.
No; the bilingual scope gets listed upfront. Deliberate gaps are fine; unplanned patchwork is not.
Through flow: content is not finished until its second-language step is done, and each language is measured separately.
Yes; patchwork pages read as carelessness. Fewer complete pages beat many half-done ones.