SEO Translations

I start with market research and local keyword research. Without understanding what people actually type into search engines in each target locale, translation is guesswork. For example, when expanding an e-commerce site into Germany, I analysed search volumes for common product terms. The literal German translation of 'wireless headphones' is 'drahtlose Kopfhörer', but that phrase has 2,000 searches per month, while 'Bluetooth-Kopfhörer' has 12,000. I would have bet on the literal version if I had not checked. Always research before you translate.

Next, I audit existing content and prioritise which pages to translate. Typically, the top 20% of pages that drive 80% of traffic get my attention first. For a SaaS client, that meant translating the homepage, pricing page, and three key feature pages. The blog archive could wait. This prioritisation saved thousands in translation costs.

Then comes the translation work. I hire a native-speaking copywriter who understands SEO, not a general translator. The difference is huge: an SEO-aware writer will naturally work long-tail keywords into headings and body copy, while a translator aims for linguistic accuracy alone. I also require them to rewrite meta titles and descriptions separately, targeting the local keyword set. For a French project, 'best CRM for small business' became 'meilleur CRM pour TPE et PME' because that is what French business owners search. I have [covered why this matters more broadly](/international-seo/) elsewhere.

On the technical side, I implement hreflang tags and use a subdirectory structure under the same domain (/fr/, /de/) to preserve link equity. Subdomains can dilute authority. I have also found that [localisation](/localization-seo/) goes beyond language — adapting offers, currency, and testimonials to the market. SEO translations that ignore cultural context will have lower conversion rates even if they rank. Finally, I test everything in Google Search Console: check that the correct language version is indexed and no soft 404s appear.

The biggest mistake I see people make is translating without [keyword research](/international-seo-keyword-research/). You end up with pages that are linguistically correct but rank for zero relevant queries. I have fixed campaigns where traffic doubled just by replacing literal translations with locally researched terms. Do not skip this step.

Skipping metadata localisation is another trap. Your meta title and description often determine whether someone clicks. If you leave them in English, or simply translate the English version literally, you miss the chance to include a compelling local keyword. I once saw a German meta description that read 'Kaufen Sie die besten Produkte' when the top-ranking competitors used 'Top-Angebote' — small change, big difference in CTR. Treat metadata as a separate optimisation task.

Automatic redirects based on IP address can break indexing entirely. Googlebot often crawls from US-based IPs, so it might never see your German or Japanese versions if you redirect based on location. Instead, let users choose via a language switcher and use hreflang to signal alternatives. I have detailed this in my guide on [multilingual SEO](/multilingual-seo/). Never rely on browser-language detection alone.

Over-relying on machine translation without human SEO review is dangerous. Machine tools have improved, but they still miss nuance. For example, translating 'bank' into Spanish as 'banco' (financial institution) is fine in one context, but if the English original meant 'river bank', you have a problem. That kind of error harms user trust and signals low quality to search engines. Use machine translation as a first draft, then have a native SEO writer edit it.

Ignoring link building for each language version is another pitfall. Ranks are built with links. If you only get links from your home market, translated pages will struggle. Plan for [international link building](/international-link-building/) as part of your roll-out. Also, watch out for local intent differences within the same language — a UK audience searches for 'nappies', not 'diapers'.

I have made several expensive mistakes with SEO translations. The first was launching a German version of my own site by translating every page literally, without any keyword research. The results were humiliating: after three months, the German site had fewer than 50 organic visits per month. I had spent money on translation and hosting for nothing. The fix was a full rewrite based on local keyword research — traffic went from 50 to 3,000 visits a month within four months. I now treat that initial project as a cautionary tale.

Another error: I used automatic browser-language detection to redirect users. I thought it was smart UX. But Googlebot's crawler had its language set to English, so it always got redirected to the English site. The German pages never got indexed. I only caught this because I noticed no German URLs in Search Console. I switched to a manual language switcher and added hreflang tags; indexing recovered within weeks. That mistake cost me half a year of visibility.

I also underestimated cultural adaptation. For a campaign targeting Japan (see my notes on [SEO Japan](/seo-japan/)), I kept the English headline structure: '5 Ways to Save Money'. That direct, listicle style works in English but feels pushy to Japanese users, who prefer softer, more informative framing. The Japanese version performed badly until I rewrote it as 'お金を節約するための5つのヒント' (5 tips for saving money) with a respectful tone. The open rate improved dramatically.

Finally, I used to think machine translation plus a brief human polish was enough. I now know that is only acceptable for low-priority pages. For any page where I want actual rankings, I invest in a native SEO copywriter. The cost difference is often 2-3x per word, but the return on that investment is 10x in organic traffic. I learned that the hard way.