What Goes Wrong When You Build an English Version of Your Site

I get asked about this often. A company translates its site, publishes it, and nothing happens.

Usually the English itself is not the problem. The problem is that the Japanese page was translated in place: same information, same order, same volume. Here is why that does not work.

The two versions need different structures

Japanese corporate sites follow a recognisable shape. They open with the company's stance and philosophy, build up credibility through track record, and arrive at what they can do for you near the end. It assumes a reader who reads in order.

English speaking readers do not read in that order. They decide from the first screen whether this is relevant to them. If they cannot tell, they do not scroll.

So translating the Japanese sequence puts the thing they came for three screens down. The failure has nothing to do with the quality of the English.

Why it needs rewriting rather than translating

Some Japanese phrasing loses its meaning entirely when translated directly.

"We stay close to our customers and propose the optimal solution" carries a certain meaning in Japanese. Rendered literally in English it describes no company in particular and no service at all. English readers read that as content free.

The reverse also happens. English sales copy translated straight into Japanese reads as pushy, and pushy reads as untrustworthy here.

What you need is not a translator but a writer who understands both audiences, producing the same substance in a form that works in each language.

What happens when machine translation ships unedited

The quality genuinely has improved. Publishing it unedited still leaves two problems.

Terminology and names. Your own product names get translated. Industry terms come out as words nobody in that industry uses. Readers notice this immediately.

Consistency. The register drifts between pages, and the same concept appears under two different words. Each sentence is correct and the whole reads as amateur.

Machine translation is an excellent first draft. The difference between publishing it raw and having a person edit it through is much larger than the cost difference.

The technical parts that get missed

hreflang. The markup where each language version points at the other. Without it, search engines may treat the two as unrelated near duplicates.

Separate URLs. Either a path like /en/ or a subdomain. Either is fine. What does not work is switching language at the same URL, because then search engines only ever index one of them.

No automatic redirect. Sending visitors to the Japanese version based on browser language means English speakers cannot reach the English site at all. Offer the switch and let the reader choose.

Layout and type. English runs longer than Japanese in most cases. Put real English text into your buttons and headings and see what breaks. Layouts sized for Japanese routinely cannot hold the English.

The most common reason nobody contacts you

There is no working way to contact you from the English site.

The Japanese version has a contact form. The English version translated the homepage and left the form in Japanese. Or the form is translated but its required fields assume a Japanese address and phone number, so an overseas visitor physically cannot submit it.

A seven digit postcode field, a phone format expecting Japanese separators, a prefecture dropdown with no other option. Each of those stops an overseas submission dead.

After you publish an English version, submit the form yourself from outside Japan. A surprising number of companies are losing every enquiry at exactly this point.

The minimum that works

You do not need to translate everything. Fewer pages usually perform better.

You need four things. What the company does. Who it is for and what it solves. Evidence of work actually done. And a contact method that works.

With those four in real English, a small site generates enquiries. Without them, fifty translated pages will not.