Multilingual website design that Google reads as one site, not duplicates

The problem with translate widgets

The cheapest way to have an English site is to bolt on a translation widget. It is also close to worthless, for three reasons:

Google does not index that translation. The content is translated in the browser after the page loads. As far as a search engine is concerned, your site is still Vietnamese only. Somebody in Singapore searching in English will never find you.

Machine translation mangles names and industry terms. Company names, product names and units of measure come out wrong in ways that make a reader doubt everything else.

Titles and descriptions are not translated at all. Which is precisely the part that appears in a search result.

How it is done properly

Each language gets its own URL: /vi/lien-he and /en/contact. These are two real pages, indexed separately, each declaring the other through reciprocal hreflang tags — which is how you tell Google "these are two versions of one thing" rather than letting it conclude you are duplicating content.

Titles, descriptions, share images and structured data are all translated, not only the visible text.

The language switcher keeps you on the same page. This sounds obvious, and yet most multilingual sites dump you back on the home page when you switch, which is exactly when people leave.

About the translation itself

We build the multilingual infrastructure; the translation has two routes.

You supply it — many companies already have English materials — and we place it correctly. Or we commission a translator, quoted per word, with your approval before anything goes live.

What we will not do is run your content through a machine and call the result an English site. For a company trying to sell to foreign buyers, ungrammatical English on the home page does more damage than having no English page at all.