Multilingual Websites: When They're Genuinely Worth It and When They're Wasted Money
Translating your website looks like a decision about reach. It's actually a decision about maintenance. The extra language isn't paid for once: it's paid for every time you change a price, publish an article or fix a sentence.
August 8, 20226 min readIn this article
- When it's worth it, no hesitation
- When it isn't worth it, however good it sounds
- If you do it: what to translate first
- URL structure: subfolder, nearly always
- hreflang: the tag that stops you competing with yourself
- Machine translation: for understanding, not for selling
- The language switcher, and the mistake of redirecting by IP
- The real cost: the first time and every time after
- Frequently asked questions
The conversation usually opens the same way: "I want the site in English too". And the question almost nobody asks first is the only one that matters — is anyone searching in English for what you sell, in the place where you can actually sell it?
Because an extra language doesn't double your reach: it doubles your work. Every price change, every new service, every article, every correction now happens twice. If the return doesn't justify that permanent weight, the translated version ages on its own and ends up making a worse impression than not having one.
When it's worth it, no hesitation
There are situations where the arithmetic is obvious. In all of them there's real, verifiable demand in the other language — not a hunch.
- Tourism and hospitality. A hotel taking foreign guests sells less for every language it lacks. Covered in hotel websites.
- Genuine export or international sales, with logistics for that country already solved.
- Professional services for expatriates: property, health, legal and immigration work in cities with a foreign community.
- A country with two languages in everyday use, where ignoring one means ignoring half the market.
- Software or a digital product, where the marginal cost of serving another country is essentially zero.
When it isn't worth it, however good it sounds
The most common case of wasted money: a local business translating into English "just in case". The predictable outcome is a version nobody visits, nobody maintains, and which two years later still shows two-year-old prices.
If you do it: what to translate first
Translating the entire site at once is the fastest way to abandon it. The strategy that works is translating what decides the purchase and leaving out what merely feeds traffic.
| Page | Priority | Why |
|---|---|---|
| Homepage, services or products, prices | High — always translate | It's where decisions happen, and what gets shared as a link |
| Contact, directions, policies | High | Without these the sale collapses at the last step |
| Case studies, portfolio, about us | Medium | Helps convince, but partly survives on images alone |
| The entire blog | Low — rarely pays for itself | Multiplies maintenance for little direct return |
URL structure: subfolder, nearly always
There are three ways to organise this and the practical difference is large. For 90% of businesses the right answer is the subfolder — `yourdomain.com/en/` — because it concentrates all authority on one domain and is the cheapest to maintain.
- Subfolder (`/en/`): one property, one certificate, shared authority. The default choice.
- Subdomain (`en.yourdomain.com`): Google treats it almost as a separate site; sensible when teams and content genuinely differ.
- Country domain (`yourdomain.de`): the strongest geographic signal, but several domains to buy, renew and rank from zero.
And one decision that gets overlooked: translate the slugs too. An English URL reading `/en/pagina-web-para-restaurantes` wastes the relevance signal and looks careless. Give each language its own readable address.
hreflang: the tag that stops you competing with yourself
Without hreflang, Google may read your versions as similar pages competing with each other, and pick the wrong one to show — or none. It's a small tag, and it's the technical piece that most often breaks on multilingual sites.
Three rules solve nearly every problem: each page declares itself and all its versions, the declarations are reciprocal (if A points to B, B must point to A), and there's an `x-default` stating where to send anyone who matches no declared language.
Machine translation: for understanding, not for selling
Machine translation has improved enormously and works perfectly well for letting someone understand your site. The problem appears when that version carries your commercial argument: selling copy depends on nuance, and a mistranslated nuance costs credibility exactly where you need it.
The practical rule: use machine translation as a starting point, and have a person review and rewrite the pages that sell. German has to read as though it were written in German, not translated from Spanish. On the blog, lightly reviewed machine translation is usually enough.
The language switcher, and the mistake of redirecting by IP
A frequent and irritating error: detecting the country by IP and redirecting automatically. It breaks the expectation of whoever followed a specific link, traps the tourist who wants to read in their own language, and confuses Google's crawler, which typically visits from the United States.
- Name languages in their own language: "Deutsch", not "German". That's how the person looking for it recognises it.
- No flags. A flag represents a country, not a language, and always excludes someone.
- Switching language keeps the page, rather than dumping you on the homepage. It's the most irritating failure.
- Suggest, don't redirect. A discreet notice offering the other version respects whoever already chose.
The real cost: the first time and every time after
The upfront budget is the easy part to estimate: professional translation of the key pages plus the technical work of structure and hreflang. It's a moderate increment on the base project — general ranges are in how much a professional website costs.
What almost never gets calculated is the recurring cost, and that's where multilingual projects collapse. Before deciding, count how many pages you maintain in a year and multiply by the number of languages. If that figure doesn't look sustainable, the honest answer is to translate fewer pages and maintain them properly. How many you need to begin with is in how many pages a website needs.
Frequently asked questions
How do I know there's genuine demand in another language?
Look at data you already have before investing: the browser language and country of your current visitors, enquiries arriving in another language by email or WhatsApp, and whether direct competitors selling to your market have already done it. Three signals agreeing beat any assumption.
Can I start with just two or three translated pages?
Yes, and it's usually the sensible move. Homepage, your main service page and contact cover most of the decision. What matters is that navigation doesn't take a visitor from a translated page to an untranslated one without warning: the unexpected language jump is what ruins the impression.
Will translating help me rank in another country?
It helps, but not on its own. Language solves relevance; the geographic signal and local authority have to be built separately. Translating into English won't rank you in the United States if nobody there links to you and you have no real presence in that market.
What should I do with an abandoned translated version?
Choose one of two paths and don't leave the middle ground. Either revive it — update prices, review the copy, fix hreflang — or retire it by redirecting those URLs to their equivalents in the main language. What can't stay is a version carrying three-year-old information, indexed and visible.