Multilingual Websites for UK Businesses: When It Pays, What It Costs, How Not to Botch It

Quick answer

Add languages when your customers demand it, not when the plugin makes it look easy: exporters selling into specific markets, businesses serving UK communities with strong first-language preferences (Welsh has legal standing in Wales for some sectors; Polish, Romanian, Urdu and others are significant markets in many towns), and e-commerce brands expanding into the EU. Done properly (real translation, per-language SEO, maintained parity), a second language runs £1,000 to £5,000 to launch and real ongoing effort. Done as a machine-translated widget, it usually damages more trust than it builds.

The decision test: three questions

1. Do you have (or want) revenue that speaks that language? Check your analytics for browser-language and geography data, your enquiry emails for language patterns, and your market plans. “It would be nice” is not a segment.
2. Can you serve customers in that language after the click? A Polish-language site feeding an English-only phone line converts badly and reviews worse. The language promise extends to quotes, support and paperwork.
3. Will you maintain it? Every new page, price change and blog post now needs two versions. Unmaintained translations drift into being wrong — which is worse than absent.

Score three yeses and multilingual is an investment; fewer, and the better spend is almost always deepening the English site; the priorities in why your website is losing you customers come first.

Translation quality tiers (and where each belongs)

The trap is uniform quality: nobody needs transcreated FAQ answers, and nobody should machine-translate their homepage promise. Budget by page importance. A typical 10-page UK SME site professionally translated lands around £800 to £2,000 per language, before build costs.

The technical side, in plain English

• Structure: subdirectories (/pl/, /cy/) are the pragmatic default on WordPress: one site, consolidated authority. Separate country domains only make sense with real in-market operations.
• hreflang tags tell Google which language version to show whom. They’re fiddly, reciprocal, and the single most common multilingual SEO defect. A competent build (WPML or Polylang on WordPress, £0 to £200/yr in licences plus setup) handles them automatically; a bolted-on widget doesn’t.
• Translate the metadata too: titles, descriptions, image alt text, URL slugs where sensible. A Polish page with English metadata competes in the wrong index badly.
• Don’t auto-redirect by IP. Offer the language switch visibly; forced redirects infuriate travellers, VPN users and Google’s crawler alike.
• Forms, emails and checkout must follow the language through — the transactional emails are where machine-translation embarrassments hide. E-commerce adds currency/VAT display questions on top; see e-commerce website costs for that layer.

Realistic total costs

On an existing WordPress site: plugin licensing and configuration £300 to £800, professional translation of a 10-page core £800 to £2,000, per-language SEO pass £200 to £500 — call it £1,500 to £3,500 per language for a proper job, plus maintenance parity forever after. Building multilingual into a new site from day one is meaningfully cheaper than retrofitting, worth flagging in the brief if expansion is on the roadmap, and one of the platform questions we scope in every custom WordPress project. For the base build costs the language layer sits on, start from what a website costs in the UK.

The maintenance reality: what “parity” means in practice

The launch is the easy half; the second language then becomes a standing item in every content workflow. A workable parity routine for a small team:

• Tier the content. Prices, legal pages, service descriptions and checkout flows must update in both languages within days; put them on a checklist triggered by any English change. Blog posts and news can lag or stay English-only; a visible “available in English” note is more honest than a stale translation.
• Batch translation monthly. Sending your translator one monthly bundle (edits + new pages) costs less per word and actually happens, unlike per-page ad-hoc requests that queue forever.
• Assign an owner. Parity fails when it’s everyone’s job. One named person runs the checklist — the same discipline as any other site maintenance.
• Watch the language split in analytics. If the second language’s traffic never materialises after 6 to 12 months of honest effort, that’s data: retire it deliberately (redirects in place) rather than letting it rot in public.

Welsh: the one language with a legal dimension

Most language decisions in the UK are commercial. Welsh is the exception, and it is routinely misunderstood by businesses on both sides of the border. The Welsh language standards made under the Welsh Language (Wales) Measure 2011 apply to public bodies (councils, the NHS in Wales, Welsh Government, many public-sector organisations), not to ordinary private companies. A plumber in Cardiff is under no statutory duty to publish a Welsh website.

Where it reaches private businesses is through contracts. Supply services to a Welsh local authority, a health board or a housing association and the standards can flow down through the procurement requirements, sometimes specifying bilingual signage, correspondence or public-facing materials. If public-sector work in Wales is on your roadmap, ask that question at tender stage rather than discovering it in a contract schedule.

Beyond obligation there is the commercial case, which is real in the parts of Wales where Welsh is the everyday language. Two rules there matter more than budget: give both languages equal visual status rather than tucking Welsh behind a small flag icon, and use a Welsh speaker rather than a machine, because Welsh mutations make automated output conspicuously wrong to anyone who reads it. Done properly it is a genuine differentiator against national competitors who have not bothered.

Briefing a translator so you don’t pay for it twice

Most translation disappointments are briefing failures rather than translator failures. Supply five things with the copy and the quality jumps without the price moving:

• A do-not-translate list. Your business name, product names, certification bodies, scheme names and anything trademarked. Left unspecified, these get helpfully translated and become unsearchable.
• A glossary of your twenty most-used terms with the meaning you intend. Trade vocabulary is where machine output and inexperienced translators fail hardest, because the same English word covers three different things depending on industry.
• Screenshots or the live URL. Translating a spreadsheet of disconnected strings produces text that is individually correct and collectively strange. Context prevents a button label being translated as a noun.
• A tone note. Formal or familiar address is a real decision in most European languages, and once it is set it must be consistent across the site, the emails and the quotes.
• Character limits for constrained elements. Navigation items, buttons and page titles need a maximum, or you will be redesigning the header around a translated menu.

Ask for a native speaker of the target language who lives in the target market, and for one reviewer other than the translator on anything customer-facing. On a 10-page site that review adds perhaps £100 to £250 and catches the errors that would otherwise be visible for years.

What breaks in the second language

The translation is rarely what fails. The build around it is. The recurring list:

• Text expansion. Translations frequently run longer than the English — often enough to wrap a button onto two lines, push a headline off a hero image or break a three-column layout. Test the longest translated string, not the average one.
• Missing characters. A brand font may not include Polish, Romanian or Welsh diacritics, so the browser silently substitutes another face mid-word. Check the font’s language coverage before, not after.
• Hard-coded English. Theme and plugin strings (“Read more”, “Add to basket”, form validation messages, cookie banners) often sit outside the translation plugin’s reach and need separate string translation. These stragglers are what make a site feel half-translated.
• Form and data formats. UK postcode validation on a form aimed at Poland will reject every real address. Date order, phone formats, address field order and required-field logic all need per-language thought.
• The documents at the end. Quote PDFs, invoices, booking confirmations and automated reminders are generated by systems the translation plugin never touches, and they are the ones customers keep.

Add every one of these to the pre-launch sweep for each language rather than checking only the English site; the general version of that sweep is the website launch checklist, and it wants running once per language, not once per site. While you are in there, note that a language switcher built as an unlabelled flag row is also a poor accessibility pattern; the underlying reasoning sits in the guide to website accessibility for UK businesses, and the fix is simply naming each language in its own language, as text.

Selling into the EU: the layer beyond words

Language is the visible part of selling abroad and usually the cheapest. The rest of the layer, for a UK business shipping to EU consumers, includes import VAT handling and whether you register for the import one-stop-shop scheme or leave customers to face charges on delivery; customs paperwork and commodity codes; EU consumer rights that differ from UK ones, including the cancellation window; the possibility of needing a representative in the EU for data protection purposes if you are targeting people there; and returns logistics that can cost more than the original postage. Take current advice on the thresholds and schemes, because these move.

The commercial point is that none of it is a translation problem, yet all of it lands on the same project. Scope it as market entry rather than as adding a language, and check that your platform can handle multi-currency pricing, per-country tax rules and per-country shipping rules before committing, a comparison worth having early, since the platforms differ sharply here, as set out in the WooCommerce and Shopify comparison. If it cannot, the honest options are a platform change as part of the e-commerce build or a narrower launch into one market done properly.

Sources & Further Reading

• Localized Versions & hreflang, Google
• Welsh Language Standards, GOV.UK
• Languages in the UK, Office for National Statistics