Targeting French and English in a Québec Google Ads account, without the language setting

We re-read Google’s help page on September 6, 2026, in English and in French. Both say the same thing: the campaign-level language setting goes away for Search campaigns, it no longer applies to Performance Max Search ads, it keeps applying to YouTube, Display, Discover and Gmail, and it has no effect on Shopping ads. Google had already scheduled this removal for late 2025 before postponing it, so check the date in your own interface when you read this.

This is not a release note. It describes an account structure that depends on no platform setting at all, because in Québec the setting being removed was doing a job it was never designed for. The full doctrine is on our bilingual advertising in Québec page; here we cover Google Ads Search and Performance Max, and the migration procedure.

What Google is removing, and what it is not

Campaign language targeting filtered the audience by the languages Google associated with a user: interface language, search history, browser preferences. It never read the language of the query itself. Someone with an English browser typing “lit en bois massif” was eligible for an English campaign and excluded from a strict French one.

After the removal, Google says it relies on the language of the ads and the landing page, combined with its understanding of the user’s known languages and of the search term, to serve the ad whose language matches. Search Engine Land notes that the scope also covers AI Max for Search campaigns.

Three things do not move. Video, Display, Discover and Gmail campaigns keep their setting. Shopping ads never depended on it: their language comes from the Merchant Center feed. And nothing changes in location targeting.

Why a Québec account is hit harder than a French one

In France, a Search campaign targets a country where one language dominates. Removing the setting changes nothing structurally, at most a marginal volume shift.

In Québec, the same geotargeting serves two official languages. The Office québécois de la langue française study on online shopping, run in fall 2023 with 3,534 respondents, found that 56.1% of consumers most often search with French keywords, 24.6% most often in English and 18.7% equally in both. In the Montréal metropolitan area, the share shopping most often in French drops to 50%, against 68% in the Québec City area. Among 18 to 34 year olds the split is close to three thirds.

In the Québec accounts we take over, the language setting rarely served to filter an audience. It served as a partition: one report per language, one budget per language, ads that do not mix. The separation rested on it because it was there. When it goes, an account whose ad groups mix keywords from both languages has no partition left.

One myth to clear on the way: this setting was never a compliance measure under the Charter of the French Language. Section 52 of the Charter concerns the existence of a French version of commercial publications, not how a platform picks its audience. The proof to produce is the coexistence of both versions, ads and pages, with French available under conditions at least as favourable. No setting replaces that proof.

The four predictable failures

In a bilingual account patched together in a hurry, this is what we expect to see in the weeks after rollout.

1. English ads on French queries, and the reverse. An ad group holding both “lit en bois massif” and “solid wood bed” with one ad per language let Google arbitrate through the setting. Without it, arbitration relies on the language match between ad and term, which often works, except when the landing page is monolingual.
2. Monolingual landing pages attached to groups that became mixed. An English ad served on an English query landing on a French page: conversion rate drops and nobody understands why, because the report aggregates both languages.
3. Per-language reporting lost, budgets drifting. If the French and English campaigns were only distinguished by the setting, the per-language cost per acquisition reading disappears. Click costs are not the same: competition on English queries includes national and US advertisers.
4. Performance Max assets mixed in one asset group. French and English headlines in the same group: Google assembles bilingual ads, unreadable for everyone, impossible to proofread.

The structure that replaces the setting

The rule fits in one sentence: every delivery unit is monolingual, named with its language, tied to a landing page in the same language.

Search: one ad group per language, strictly monolingual ads

Two campaigns per purchase intent, one French and one English, each with keywords, responsive search ads and extensions in a single language. Headlines and descriptions contain no element of the other language, not even a translated collection name. The brand name stays as is in both.

Negative keywords do the work the setting half-did: the French campaign’s negative list receives the English terms that triggered it, and the reverse. That is a weekly search terms review, per campaign, for at least four weeks after rollout.

One landing page per language, never an automatic switch

Each ad points to the URL in its own language. No page that detects location or browser to switch language: an English speaker in Laval and a French speaker in Toronto both exist, and detection gets both wrong. Both versions carry reciprocal fr-CA and en-CA hreflang tags, which also helps Google understand which page belongs to which ad.

Naming: the language prefix

Here is the convention we apply. Nothing sophisticated about it; its only virtue is that per-language reporting becomes a filter on the name.

An account label, lang:fr and lang:en, duplicates the prefix for cross-reports and automated rules.

Performance Max: two asset groups or two campaigns

One asset group per language is the minimum: no text, video or image carrying text crosses the line. Audience signals and customer lists are also split when a language field exists in your database.

Two campaigns rather than two asset groups when either condition holds: profitability targets diverge (average order value or margin differ by language, which happens more often than expected), or one of the two markets is too small to exit learning and deserves a protected budget. Otherwise one campaign with two asset groups is enough and keeps learning pooled.

The language checkpoint that remains mandatory

Removing the setting removes nothing from the Charter. Section 52 requires commercial publications, including on a website and on social media, to exist in French, and the Regulation respecting the language of commerce and business allows another language if the French version is available under conditions at least as favourable.

What an advertiser must be able to show, with no language setting: that every English ad has a French twin live, that every English landing page has its French version, and that French is not relegated. We keep a dated screenshot of both versions at every launch. A ten-minute proof worth more than a vanished setting.

One question we have seen settled nowhere: does a geotargeted paid ad fall under commercial advertising in the sense of section 58 (marked predominance of French) or under a commercial publication in the sense of section 52 (parity of access)? We apply the stricter requirement and recommend having that choice validated by legal counsel. This is not legal advice.

Seven-day execution plan

Days 1 and 2: inventory. Export the list of Search and Performance Max campaigns with their current language setting. Flag those where the setting is the only thing separating French from English.

Days 3 and 4: ad groups and assets. For each flagged campaign, check that every ad group is monolingual: keywords, ads, extensions. Split what is not. In Performance Max, open each asset group and move the other language’s assets into a dedicated group.

Day 5: landing pages. Open every final URL and confirm its language. Fix automatic switching if the site does it. Check hreflang.

Day 6: naming and labels. Apply the language prefix and the labels. This is what makes tracking possible.

Day 7: tracking grid. Create two views of the search terms report, filtered by language prefix, to review weekly for four weeks. Track per language: share of terms in the other language, click-through rate, conversion rate, cost per acquisition. A drift above 10% of wrong-language terms flags a group that is still mixed.

This plan needs no budget. It needs a day of work on an average account, and it makes the account readable per language for the first time, which the removed setting never allowed cleanly. If you would rather we run the inventory, our Google Ads in Montréal page describes how we operate Québec accounts, and a review starts precisely there.