If you manage campaigns in bilingual markets or run international accounts, your account architecture is facing a forced restructuring. Starting September 2026, Google Ads is deprecating campaign-level Google Ads language targeting across Search and Performance Max campaigns. For over a decade, setting explicit language parameters was the default safeguard against serving English ad copy to Spanish-speaking searchers, or burning German budget on French search queries in multilingual hubs like Switzerland and Belgium.
Google’s stated reasoning follows its standard playbook: machine learning now supposedly detects user language intent dynamically based on query syntax, recent browsing history, OS settings, and location signals. But removing deterministic campaign controls shifts the entire burden of traffic filtering onto your keyword match types, negative keyword lists, and landing page routing. If you leave your multi-language campaign setup on autopilot heading into Q4, your ad relevance will drop, conversion rates will degrade, and Smart Bidding algorithms will misallocate budget across mismatched language assets.
Why Google Is Sunsetting Campaign-Level Language Settings
Google's legacy language targeting functioned on a composite model: the platform matched ads based on a user's interface language setting combined with the detected language of their search query. If an advertiser selected "Spanish" under campaign settings, Google served ads only when the user’s profile language or active query signaled Spanish fluency.
That system is being replaced by fully automated query-intent matching. Under the new framework, Google Ads determines language delivery primarily through the semantic language of the query itself and the language used in your ad assets.
Legacy Model:
[Campaign Language: Spanish] + [Spanish Keyword] + [Spanish User Interface] = Ad Serves
New Model (Sept 2026):
[Query Detected as Spanish] + [System Matches to Any Active Ad Asset] = Ad Serves
The issue is that automated language detection breaks down in real-world advertising scenarios. Words with identical spelling across multiple languages (homographs), technical jargon, brand names, and transliterated queries routinely confuse AI language classifiers. When the platform misclassifies a query, it will now match to whatever ad group has the highest ad rank, regardless of the language your ad copy is written in.
The Three Places Where Budget Will Leak First
Without a campaign-level language filter acting as a hard boundary, budget waste will concentrate in three distinct areas of your account.
1. Homographs and Shared-Vocabulary Search Terms
Dozens of languages share identical words with completely different intents. In English, "gift" means a present; in German, "Gift" means poison. "Real" is an English adjective and a Spanish noun for currency or royal status. "Boutique" spans English, French, and half a dozen other European languages.
When you run broad match keywords without a strict campaign language fence, Google’s query expansion matches these identical tokens across languages. In accounts spending $50,000+ per month across Europe or North America, homograph leakage routinely burns 4% to 8% of total ad spend on search terms that have zero conversion intent for the actual product offered.
2. Ad Relevance and Expected CTR Degradation
Quality Score relies heavily on Ad Relevance and Expected CTR. When a user searching in French triggers an English ad group because the keyword matched a universal brand name or broad match variation, that user will not click.
If they do click out of curiosity, they land on an English page they cannot read, creating an immediate bounce.
Mismatched Query Language → Sub-1% CTR → Dropped Ad Relevance → Lower Quality Score → 25-40% Higher CPCs
Smart Bidding will eventually recognize that the query does not convert and bid down, but only after wasting dozens of expensive clicks across every active ad group.
3. Performance Max Asset Group Cannibalization
Performance Max already struggles with multi-language asset allocation. When you run a single PMax campaign with multiple asset groups targeting different languages in the same geographic region, Google's algorithm frequently serves English text headlines alongside Spanish image extensions, or delivers English search text to a user whose query was entirely in Portuguese.
Once language targeting is removed at the campaign level, PMax will freely match any asset group to any query it deems relevant based solely on predicted conversion value. If one language group has higher conversion volume—typically English—the algorithm will starve your localized asset groups of impressions, effectively destroying your localized conversion funnels.
How to Restructure Your Multi-Language Accounts Before September
Waiting until the September rollout to adjust your account setup will trigger volatile bidding swings right before peak Q4 seasonality. You need to implement structural guardrails now.
Isolate Campaigns by Dedicated Negative Language Dictionaries
Because you will no longer have a settings toggle to exclude languages, you must build robust negative keyword lists containing common stop words and prepositions for every language you do not want to serve in that campaign.
If you run an English-only campaign in Canada or the US Hispanic market, apply a shared negative keyword list containing high-frequency non-English stop words:
- Spanish Stop Words Negative List: de, la, que, el, en, y, a, los, se, del, las, por, un, para, con, no, una, su, al, lo, como, más, pero, sus, le, ya, o, fue, este, ha, sí, porque, esta, cuando, muy, sin, sobre, también, me, hasta, hay, donde, quien, desde, todo, nos, durante, todos, uno, les, ni, contra, otros, ese, eso, ante, ellos, e, esto, mí, antes, algunos, qué, unos, yo, otro, otras, otra, él, tanto, esa, estos, mucho, quienes, nada, muchos, cual, sea, poco, ella, mayor, mucha, junto, zon, bien
- French Stop Words Negative List: de, la, le, et, les, des, en, un, du, une, que, est, pour, qui, dans, a, par, plus, pas, au, sur, ne, avec, se, son, sont, aux, ou, ses, ont, sa, mais, comme, nous, tout, il, c, fait, vous, ce, été
Attaching these lists at the campaign level enforces the deterministic boundary that Google is removing from the settings panel.
Restructure Landing Page Signals and Hreflang Tags
Google’s crawler scans your final URLs to help its machine learning models classify the intended language of your ads. If your landing pages feature mixed-language navigation bars, untranslated footers, or missing hreflang annotations, Google’s automated classifiers will misassign your ad groups.
Ensure that:
- Every localized ad group points exclusively to a fully localized URL path (e.g.,
/fr/or/es/), never to a generic root domain that relies on IP-based auto-redirects. - All target URLs have valid
hreflangtags pointing directly to the alternate language versions. - The on-page content, including metadata, title tags, and schema markup, is 100% translated into the target language.
Audit Account Performance for Hidden Language Bleed
Most advertisers are unaware of how much multi-lingual crossover already occurs in their accounts under the current settings. During a standard Gromerce audit, one of the first structural flaws we look for is query-language mismatch across international campaign setups. Running an audit on your search term data today allows you to establish a baseline of non-target language spend before the platform-level control is permanently removed.
What to Do This Week
Do not wait for Google to flip the switch in September. Take these three actions across your active Search and Performance Max campaigns this week:
- Pull a 90-day Search Term report and filter by script and common non-target language characters. Look for accented characters (
é,à,ñ,ü,ç) inside campaigns intended strictly for standard English. Calculate the percentage of spend allocated to non-target queries. - Build and apply Shared Negative Keyword Lists for your top 3 non-target languages. Populate each list with at least 50 high-frequency prepositions and auxiliary verbs from languages you do not sell to in those campaigns.
- Split multi-language PMax campaigns into dedicated single-language campaigns. If you are running multiple language asset groups inside a single PMax container, break them out into separate campaigns with strictly mapped localized product feeds and dedicated negative brand/language lists.
Sources: Search Engine Journal
