Website Localisation

What sets it apart most from other translation work is that the project never ends. A site is updated continuously; new products, new campaigns and new blog posts keep arriving. Localisation therefore has to be set up as a flow rather than as a delivery.

Internationalisation comes first

If a site is not technically ready to be multilingual, the result will come out flawed however good the translation. Text embedded in the code, slogans baked into image files, variables joined together in mid-sentence and fixed-width buttons are the typical obstacles. Putting these right is development work, not translation work, and it has to happen before translation begins.

Text length is a separate matter. Moving from Turkish into German, text can grow noticeably; if the interface is not ready for that expansion, button labels are cut off. Building flexibility into the design costs less than dozens of corrections afterwards.

Content layers do not all call for the same method

Not all the text on a site is equal. A homepage headline and a product description, a returns policy and an error message each call for a different approach. Using the budget efficiently means separating the layers.

Layer	Example	Method	Update frequency
Brand and campaign copy	Homepage headline, slogan	Rewriting, cultural adaptation	Low, but high impact
Product and category text	Descriptions, specification lists	Translation against a termbase	High
Interface text	Buttons, menus, error messages	Character-limited translation	Low
Legal texts	Privacy, returns, terms of use	Review against target-market regulation	Low, but high risk
Blog and support content	Guides, frequently asked questions	Search-led translation	Continuous

Search visibility cannot be thought about apart from translation

A user in the target market may not search for your product with the word you use for it. A directly translated heading can be grammatically flawless and still be a phrase nobody ever types into a search box. Page titles and category names should therefore be chosen according to how people actually search in the target language.

On the technical side, the language versions need to be linked to one another correctly. Each version should have its own address, the versions should point to each other, and the search engine should be told plainly which version serves which language. Where those links are missing, language versions can be read as copies of one another.

The third issue is redirecting visitors automatically by language. Sites that force a redirect based on browser language take the choice out of the user’s hands and make it harder for search engines to crawl every version. The right approach is to suggest a language while leaving the user free to change it.

Where does cultural adaptation begin?

The least visible but most telling side of localisation is everything outside the text. The order of the fields in an address form, the format of a telephone number, the arrangement of first name and surname, which day the week starts on, units of measurement and size charts: together these decide whether a user feels the site is their own.

Images belong here too. A photograph that looks natural in one market may look beside the point, or simply wrong, in another. Colours and symbols likewise do not carry the same associations in every culture. These decisions belong to the marketing team rather than the translator; they are, however, questions that need raising during localisation.

Payment and delivery expectations are the most concrete difference of all. A payment method that is standard in one market may go entirely unused in another. However well the site is translated, conversion stays low if it offers no familiar way to pay.

Measurement and improvement after launch

The success of localisation is measured by behaviour in the target market, not by the number of words delivered. The indicators to watch are clear: bounce rate by language version, add-to-basket rate, the words typed into the search box and the substance of the questions reaching the support team.

On-site search logs are particularly valuable. The words users look for but fail to find show you directly where your choice of terms does not match the target market. That data is the most reliable input for a revision of terminology a few weeks after the first translation goes live.

A sound localisation plan therefore does not end on launch day; it builds in the first period of measurement, and a round of corrections based on it, from the outset.

Where should you start?

Translating a whole site in one go is both the most expensive and the riskiest route. The more efficient approach is to start with a narrow path that earns revenue: the homepage, the best-selling category, the product page template, the basket and checkout steps, and the returns and delivery terms. This core set covers the entire purchase journey and usually accounts for a small share of the text on the site.

Once that narrow path is live, the data it produces shapes the second wave: which categories draw interest, which questions the site fails to answer, which content is searched for but not found. Budget is then allocated on real demand rather than guesswork. Blog and support content usually earns its place in this second wave.

The staged approach brings one more benefit: the first wave lets the translation team learn the brand’s language and puts the technical groundwork to the test under real conditions. Whatever problems surface are resolved within a small body of content; had they come to light after the whole site was translated, the cost of putting them right would have multiplied.

How the work is organised

1. Inventory. The total body of text on the site is established and the amount of content in each layer measured.
2. Technical preparation. Text is separated from the code, language files are created and the address structure is settled.
3. Style and terminology guide. The brand’s tone of voice in the target language, its forms of address and its fixed terms are written down.
4. First wave of translation. Work starts with the pages that earn revenue; the whole site is not translated at once.
5. Pre-launch check. Translations are reviewed in place, on the actual interface; text that overflows or reads oddly out of context is corrected.
6. Continuity. A regular rhythm is set up for handing over new content, so that a project need not be opened from scratch each time.

What determines the price?

The first factor in the cost of localisation is the real volume of text to be translated, and that volume is usually larger than expected, because the product database gets overlooked. The second is how the text is spread across the layers: marketing copy that has to be rewritten is priced differently from product descriptions.

The third is continuity. In a continuous-flow model, the unit cost of new content falls as the translation memory grows. You can see which languages we work in on the language pages and the pricing tiers on the pricing page, and share your site inventory through the quote form.

Situations that come up often

• Machine translation plugins. They make for a quick start but carry serious risk in legal texts and product specifications; those two layers at least should be checked by a person.
• Half-finished languages. A site whose menus are translated but whose checkout is still in Turkish loses more trust than a site never translated at all.
• Text inside images. The wording on campaign banners often escapes the inventory and is noticed only on launch day.
• Customer service. Once the site is translated, the questions arrive in that language too; projects where the support side has not been prepared seize up in the first month.