Content Localisation

The Spectrum from Translation to Transcreation

Machine translation — fast, near-free, adequate for internal comprehension and low-stakes content. Visibly non-native for anything customer-facing.

Professional translation — accurate and fluent, but constrained by the source text's structure and cultural assumptions.

Localisation — adapts examples, currency, units, imagery, references and regulatory detail alongside the language.

Transcreation — recreates the intent in the target market, potentially departing substantially from the source. Necessary for campaign lines, humour and wordplay, which do not survive translation.

Match the level to the content. Product documentation needs accuracy; a brand campaign needs transcreation.

What Localisation Actually Changes

• Language, including regional variant — Spanish for Mexico is not Spanish for Spain
• Currency, price points and payment methods
• Date, number and address formats
• Examples, case studies and named references that mean nothing locally
• Imagery reflecting local people and environments
• Legal and regulatory disclosures
• Colour and symbol associations, which vary meaningfully

SEO Implications

Localised content needs its own keyword research. Direct translations of source keywords frequently miss how the market actually searches, and search volume distributes differently across synonyms.

Technically, each language version needs its own indexable URL with correct hreflang annotation. Machine-translated pages published at scale are a known low-quality pattern and are treated as such.

Workflow and Governance

A workable process: define what must stay globally consistent, produce source content in a translation-friendly way, use a translation memory to keep terminology consistent and cost down, and have a local reviewer sign off before publication.

The local reviewer is the step most often cut and the one that catches the errors that matter — the phrasing that is technically correct and culturally wrong.

What It Costs, and Where the Cost Actually Sits

Localisation budgets are usually built around translation, which is the cheapest part and getting cheaper. The expensive parts are the ones nobody lists.

Review by someone who knows the market is the line that matters. A translation is correct; a market reviewer tells you that the correct translation is off-register, that the example means nothing locally, or that the claim you make routinely at home is not made in that market. This is a recurring cost per piece, not a one-off.

Visual and layout rework. Text length changes substantially between languages. Layouts built to one language's word length break in another, and the fix is design time rather than translation time.

Ongoing maintenance is the cost that ends most localisation programmes. Every update to the source now has to propagate to every locale. A programme that localises a hundred pages into four languages has not created four hundred pages — it has created a maintenance obligation on five hundred, forever.

Localise less, deeper. A small number of properly maintained locales beats a large number of decaying ones.

The Signals That Tell You It Is Working

Localisation is easy to do and hard to evaluate, because the obvious metric — traffic in the target market — moves for reasons that have nothing to do with quality.

More useful signals, in rough order of reliability:

• Engaged time on localised pages versus the source-language equivalent in the same market. If local visitors spend longer on the English page, the localisation is worse than nothing and is actively getting in the way.
• Conversion rate by locale, compared against itself over time, not against other locales — markets differ for reasons localisation cannot fix.
• Search visibility for terms people in that market actually use, which is the test of whether you localised the keyword research or only the copy. This is the most common gap: the content is translated and the topic was chosen for a different market.
• Support contacts in-language, which rise when localisation works and reveals demand the organisation may not be staffed for.

That last one is worth anticipating. Localised marketing with unlocalised support produces a worse customer experience than no localisation at all.