Google Ads Launches AI Campaign Localization Beta for 10 Search Ad Languages

What the Tool Actually Does

The design principle is separation. The original campaign is never modified. The localized version is a fresh entity with its own performance history starting from zero, its own campaign name, and its own settings. That matters for advertisers who want to test a new market without disturbing a working campaign.

Four components are handled by the tool:

• Text assets and targeting: Headlines, descriptions, sitelinks, callouts, and keywords are all translated. The inclusion of keywords is significant—it means the targeting layer changes, not just the visible copy.
• Images: Where images carry text overlays, the AI removes the original text (in-painting) and moves the translated message into separate text assets. The translated words are not redrawn onto the image.
• Landing pages: Advertisers can embed a Google-supplied JavaScript snippet that translates page text dynamically when a user clicks an ad. Alternatively, URLs can be updated manually or via a bulk find-and-replace rule.
• Budgets and bids: These are copied exactly, without currency conversion. A $100 USD budget becomes a $100 USD budget in the localized campaign, regardless of the target market.

That last point deserves attention. Cost-per-click levels, auction dynamics, and purchasing power vary considerably between the United States and, say, Poland or Brazil. The tool copies the numbers; the advertiser decides whether those numbers make sense.

The Five-Step Workflow

Google describes the process in five steps, and the structure is straightforward.

Access comes through three routes: selecting an eligible campaign in the Campaigns table, opening Localization from the Tools menu, or acting on an optimization recommendation that Google surfaces when it identifies a localization opportunity.

Setup involves naming the new campaign, selecting the target language, setting the location, and entering a budget. A free-form field accepts global instructions—Google’s examples include “Do not translate the brand name” and “Keep the tone formal.” The help page gives no character limit for this field and does not specify how strictly the model follows it.

Review is designed to be accessible without knowledge of the target language. A four-column table shows the original asset, the translated asset, a back-translation into English, and translation notes. Each translated asset can be edited inline. The back-translation is the practical check for whether meaning has survived the model.

Landing page strategy is where the tool reaches outside the ad account. The JavaScript snippet, which Google refers to both as a “provided JavaScript snippet” and as “the Google tag,” only needs to be installed once per account. It translates page text at load time when a user arrives from an ad. The review table in step three covers ad assets; it does not cover what the script produces on the page.

Publishing defaults to a paused state, giving the advertiser time for a final review before the campaign goes live.

For readers tracking broader changes in Search campaigns, this workflow fits into Google’s wider push toward more AI-assisted campaign setup.

The Landing Page Script: A Closer Look

The JavaScript snippet is worth examining separately, because it introduces a new piece of code running on the advertiser’s website. The help page does not describe how translated text is served, whether it is cached, how it interacts with consent banners, or whether translated page content is visible to search crawlers. It also does not address whether page text processed by the script is stored or reused by Google.

For advertisers, the practical gap the script addresses is real: a localized ad pointing to an English-language page is a known conversion problem. The script closes that gap without requiring the advertiser to build separate translated pages. The tradeoff is that the translated page is generated per visit and is not part of the five-step review flow.

Supported Languages and Eligibility

The tool currently supports Search campaigns only, with English (US) as the required source language. The 10 target languages are:

• Dutch
• French
• French (Canada)
• German
• Italian
• Polish
• Portuguese (Brazil)
• Portuguese (Portugal/European)
• Spanish (Castilian/Spain)
• Spanish (Latin American & Caribbean)

Counted by base language, that is seven languages, with regional variants for French, Portuguese, and Spanish. No Asian, Nordic, or Middle Eastern language is included. British English is not available as either a source or a target.

Eligibility rules are narrow but firm. The source campaign must be active or paused—removed, draft, experimental, and shadow campaigns are excluded. Every asset and keyword in the source campaign must be in English (US); a US campaign mixing English and Spanish keywords to reach bilingual users cannot be localized. Election-verified advertisers are excluded entirely, consistent with Google’s existing disclosure requirements for synthetic or digitally altered content in political advertising.

Why This Tool Arrives Now

The timing is not coincidental. In August 2026, Google confirmed it would remove campaign-level language targeting from Search and Performance Max campaigns—a change originally promised for the end of 2025 that arrived nine months late. With that setting gone, the language of the ad creative and the landing page has become one of the primary signals Google uses to match ads to users. That gives advertisers a structural reason to run separate campaigns per language rather than relying on a single campaign with language targeting applied.

The localization tool produces those separate campaigns. Google’s earlier statement that it “does not translate ad creatives” described the ad-serving layer; this tool operates at the campaign-building stage, where the advertiser initiates the translation and reviews the output before publishing.

The contrast with Meta’s approach is also worth noting. Reports surfaced on October 22, 2026, that Meta’s AI had been translating ad copy and adding sitelinks in live campaigns without advertiser requests, while Meta’s revised Commercial Terms make advertisers solely responsible for changes its AI features make. Google’s tool is opt-in, writes into a new campaign rather than modifying an existing one, and defaults to paused. The responsibility question is not fully resolved—the help page does not address who is accountable for translation errors or how translated assets are checked against ad policies in each target market—but the structural design gives the advertiser more control than the Meta model described in those reports.

This rollout also sits alongside other Google Ads AI changes, including newer reporting and analytics insights.

What Remains Open

Several practical questions are unanswered. The help page is undated and carries no rollout information, so it is unclear what share of accounts currently has access. The beta label means the feature can change. The help page states that Performance Max support would begin in September 2026; no confirmation of that has appeared in the available documentation. The page sets no limit on the number of localized campaigns per account, provides no detail on how translated keywords are matched to queries in the target language, and offers no performance data from beta testers.

Google describes the benefit in terms of speed: replacing “extended periods” of manual translation and spreadsheet uploads with a few clicks. That is a claim about setup time, not about results.

The Practical Takeaway

For advertisers running US English Search campaigns who want to enter Western European or Latin American markets, the tool meaningfully reduces the mechanical work of campaign setup. The translation covers assets and keywords, the review step provides a back-translation check, and the paused default prevents accidental spend.

The decisions that remain with the advertiser are the ones that matter most commercially: whether the budget figure makes sense in a different currency environment, whether the bids are competitive in a different auction, whether the translated copy meets local legal and cultural standards, and whether a JavaScript snippet on the landing page is an acceptable substitute for purpose-built localized pages. The tool handles the scaffolding. The judgment calls are still yours.