Localization is the work of adapting a product, an advertisement or a piece of content so that it fits the language, cultural expectations, formats and legal rules of a specific market. The World Wide Web Consortium (W3C) defines it as adaptation "to meet the language, cultural and other requirements of a specific target market (a locale)." Translation is one part of it. The rest covers currency, dates, units, imagery, colours, legal wording, references that do not travel, and the landing page an ad points to. Localization exists because a campaign written for one country rarely works unaltered in another, and because buyers overwhelmingly prefer to read about products in their own language.

How localization works in an ad campaign

The unit of localization is the locale, a combination of language and region. French for France and French for Canada are separate locales, as are Spanish for Spain and Spanish for Latin America. Standards express locales as language tags such as fr-FR or pt-BR, codes that recur in ad platforms, product feeds and hreflang annotations.

The W3C's list of what localization can touch runs from date and currency formats to symbols, colours, legal requirements and text that may be "subject to misinterpretation or viewed as insensitive" in a given culture. In advertising, a Search campaign has headlines, descriptions, sitelinks, callouts and keywords. A display or social campaign adds images with text burned into them, video voice-over and subtitles. Shopping campaigns pull prices, shipping terms and product titles from feeds that must carry local currency and units.

A typical localization workflow runs in five stages:

  • Source preparation. Strings are extracted from the master creative, with notes on brand names and claims that must not change.
  • Translation. A human linguist, a machine translation engine or both produce the target-language text, usually drawing on a translation memory and a glossary of approved terms.
  • Adaptation. Copy is rewritten where literal translation fails, prices and offers are swapped for local ones, images are changed, and character limits are rechecked. German compounds and Finnish case endings regularly overflow a 30-character headline field that English fits comfortably.
  • Review. A native-speaking reviewer checks tone and accuracy. Back-translation, rendering the target text back into the source language, is often used so that a non-speaker can see what the new version says.
  • Compliance and launch. Legal checks run against local rules before the localized version goes live, often as a separate campaign.

Keyword localization is its own discipline, because a literal translation can be a term nobody searches for. Amazon's global keyword tool, according to its international Sponsored Products guide, "automatically translates your keywords to the best-matching keywords" in the local marketplace's default language rather than offering a word-for-word rendering.

Brand teams, agencies and language service providers do the work on the buy side. Increasingly, the platforms do it themselves. Google's AI campaign localization tool, in beta since October 2026, takes an English (US) Search campaign and produces a separate campaign in one of 10 target languages, including separate variants for Canadian French, Brazilian Portuguese and Latin American Spanish. According to the Google Ads Help Center, it translates headlines, descriptions, sitelinks, callouts and keywords, removes text from images "(in-paint) to provide a clean image" and adds the translated message as text assets. Its review table shows each original asset, the translation, a back-translation into English and translation notes. The new campaign is published paused by default. Budgets and bids carry over unchanged: "a $100 USD budget becomes a $100 USD budget in the localized campaign," the help page states, because the system does no currency conversion.

Origin and evolution

Localization as a named practice came out of the software industry, where companies shipping products abroad needed to separate translatable text from code. The Localization Industry Standards Association (LISA), a Swiss-based trade body, was founded in 1990 to set standards for that work. Its terminology exchange format, TBX, became ISO 30042:2008. LISA closed on February 28, 2011, and an Industry Specification Group at the European Telecommunications Standards Institute (ETSI) took over several of its work items.

The web extended localization from boxed software to sites and then to advertising. Google published new markup for multilingual content in December 2011, the basis of today's hreflang annotations. In paid search, advertisers expanding abroad built separate campaigns per country and language, managed through spreadsheets and agency translation queues.

Generative AI changed the economics. Nimdzi, a language industry analyst, estimated the language services market at $71.7 billion in 2024 and projected $75.7 billion for 2025, with machine translation and post-editing driving the largest revenue gains. It also cut its 2029 forecast from more than $100 billion to $92.3 billion as per-word prices fall. Ad platforms moved into that gap. Meta announced AI dubbing with voice synthesis and lip sync for advertisers at Advertising Week in October 2025, initially between English and Spanish, according to Slator. Search Engine Land reported on September 18, 2026 that Google Ads had opened Video Ads Dubbing in Asset Studio in 33 languages and locales for select advertisers.

Why localization matters for marketers

The commercial case rests on consumer preference. A July 2020 survey by CSA Research of 8,709 consumers in 29 countries found that 76% prefer to buy products with information in their own language and 40% will never buy from websites in other languages. "Localization improves customer experience and increases engagement in the brand dialogue," Donald DePalma, CSA Research's chief research officer, said at the time.

Regulation sets a floor under that preference. France's Toubon law of August 4, 1994 applies its French-language requirements to "toute publicité écrite, parlée ou audiovisuelle," and requires that a French version be "aussi lisible, audible ou intelligible" as any foreign-language one. A French culture ministry review of 19,323 advertisements between January and April 2018 found 4.8% non-compliant; 88.7% of those breaches were missing translations. Quebec tightened its rules under Bill 96 on June 1, 2025: commercial advertising must be in French, other languages are permitted only if French is "markedly predominant," and on signs the French text must be at least twice the size of the other language, according to DLA Piper.

Platform design pushes the same way. Google removed the campaign-level language setting from Search and Performance Max in September 2026. Platforms now infer language from browser settings and behaviour, which makes the language of the creative a stronger signal.

Limitations, criticism and open disputes

Localization failures are cheap to make and expensive to fix. HSBC spent a reported $10 million in 2009 rebranding its private bank after its "Assume Nothing" slogan was mistranslated as "Do Nothing" in some countries, according to WealthBriefing, citing the Financial Times. Automation reproduces that failure at scale: text that is grammatically correct and commercially wrong.

Machine translation narrows the cost gap without closing the quality gap. Back-translation, which Google's tool uses as its review aid, shows what a sentence means but not whether it sounds natural, uses the right register or breaks a local advertising rule. The Google tool's help page, as PPC Land's review of it found, does not say who bears responsibility for translation errors, does not convert budgets or bids, and does not explain how its landing page translation script handles caching, consent banners or crawler visibility. Meta's revised advertising terms, effective October 30, 2026, make advertisers "solely responsible" for what its AI features create or change, including features Meta switches on itself; one agency cited in that coverage pulled ads that had language translations added without a request.

How much quality matters is disputed. The same CSA survey found 65% prefer content in their language even if it is poor quality, which suggests cheap translation may suffice for discovery while regulated claims still demand careful work. Vendor-reported results add little clarity. Google cited an 86% increase in click-through rate and a 75% fall in cost per click for dubbed video, figures Search Engine Land described as "Google-provided examples rather than a guarantee of performance."

Localization versus adjacent terms

  • Translation converts text from one language to another. Localization includes translation and adds currency, imagery, formats, legal wording and cultural fit.
  • Internationalization (i18n) is the upstream engineering that makes localization (l10n) possible: separating text from code, supporting different scripts and leaving room for longer strings. The W3C defines it as design "that enables easy localization." The numerals count the letters between the first and last characters of each word.
  • Transcreation is the most interpretive end of localization, where a slogan or concept is rewritten from scratch for a market rather than translated, usually by copywriters.
  • Language targeting decides who sees an ad based on inferred language. Localization decides what language and content the ad contains.

Recent developments

Google's localization tool opened more widely in late October 2026. PPC Land reported today that the Google Ads AI tool copies Search campaigns into 10 new languages, after PPC News Feed first surfaced the help page on October 5 and reported its availability on October 23. The help page says more languages will follow by the end of 2026 and that Performance Max support was planned to start in September 2026; no confirmation of the Performance Max rollout has appeared. Video is moving the same way. YouTube auto dubbing passed 6 million daily viewers and expanded to 27 languages, and Google began adding AI voice-overs to silent Performance Max videos by default in March 2026. SEGA reported a 44% lower CPM using Bannerflow's dynamic social ads across seven markets in August 2026, a self-reported figure with no stated baseline.

Timeline

  • 1990 - The Localization Industry Standards Association is founded in Switzerland.
  • August 4, 1994 - France adopts the Toubon law, applying French-language requirements to written, spoken and audiovisual advertising.
  • 2008 - LISA's TermBase eXchange format is published as ISO 30042:2008.
  • February 10, 2009 - HSBC's private bank rebrand, reported at $10 million, follows the "Assume Nothing" mistranslation.
  • February 28, 2011 - LISA ceases operations; ETSI takes over several standards work items.
  • December 2011 - Google introduces markup for multilingual content, the basis of hreflang annotations.
  • July 7, 2020 - CSA Research publishes its 29-country survey: 76% prefer buying in their own language.
  • January 16, 2021 - Amazon adds translations of recommended keywords to the Amazon Advertising API.
  • June 1, 2025 - Quebec's Bill 96 rules on trademarks, signage and advertising take full effect.
  • October 2025 - Meta announces AI dubbing with lip sync for advertisers at Advertising Week, starting with English and Spanish.
  • September 2026 - Google removes campaign-level language targeting from Search and Performance Max.
  • September 18, 2026 - Google Ads Video Ads Dubbing in Asset Studio reported in 33 languages and locales.
  • October 5, 2026 - Google's AI campaign localization tool for Search surfaces in a Help Center page.
  • October 30, 2026 - Meta's revised terms making advertisers solely responsible for AI edits take effect.

Summary

Who: Brands, agencies and language service providers on the buy side; Google, Meta, Amazon and Microsoft as platforms that increasingly localize ads automatically; standards bodies such as the W3C and, until 2011, LISA; and regulators in France, Quebec and elsewhere that set language rules for advertising.

What: The adaptation of ads, products and content to the language, culture, formats and legal requirements of a specific locale, covering translation, currency, imagery, keywords, landing pages and compliance.

When: The practice took shape in the software industry, with LISA founded in 1990. It moved to the web in the 2000s, with Google's multilingual markup arriving in December 2011, and into ad platforms as generative AI tools from Meta (October 2025) and Google (September and October 2026).

Where: In every market where an advertiser sells outside its home language, across Search, social, video, retail media and Shopping feeds, and on the landing pages ads point to.

Why: Most consumers prefer to buy in their own language, some laws require it, and the removal of language targeting settings has made the language of the creative itself a stronger signal for platforms.