Bilingual campaign operations for Orleans, where half the audience searches in French

It is the most common shortcut here and it reliably underperforms. Translation gives you the words people would use if they were writing a sentence, not the words they type into a search box, and it misses the accented and unaccented variants, the borrowed English terms that are in genuine use in this market, and the longer query patterns French searchers tend to produce. Build the French list from French search data and from what your own French-speaking customers actually say, then check it against the English list for gaps rather than deriving it from one.

No, and this is the single most expensive misunderstanding in a bilingual account. It targets people based on their interface and content language settings, not on the language of the individual query. So a user whose browser and account are set to French is in your French audience whatever they happen to type, and an English-interface user typing a French phrase is not. The practical consequence is that language targeting is a coarse audience filter, and the thing that actually matches a French query is a French keyword.

Its own campaign, in almost every case. Budget is set at the campaign level, so sharing one means the language with more volume quietly absorbs the spend and nobody has decided that. Negatives applied at the campaign level stay where they belong instead of leaking across languages. And the reporting splits for free, which matters more than it sounds, because a split that requires a filter is a split nobody applies. The cost is a slightly larger account, which is a good trade.

It is a decision rather than a fault, and it should be made deliberately. Ottawa and Gatineau are one metropolitan area across a provincial boundary, so a radius that makes sense on a map crosses into a jurisdiction with different consumer rules and different French-language obligations for businesses operating there. If you genuinely serve that side, target it on purpose and account for what selling there requires. If you do not, exclude it, because the alternative is paying for clicks you cannot convert and diluting the location data you use to make the next decision.

Not before checking three things, because in a badly built bilingual account the French numbers are usually measuring the account rather than the audience. Are the French ads landing on French pages, or on English ones. Is the French campaign sharing a budget it keeps losing. Are the conversions on the French side even being tracked, or does the French version of the form sit on a page the tag never fired on. Those three explain most of the gap. If the French numbers survive all three, then it is a real result and worth acting on.

Report the same small set of metrics twice rather than reporting twice as many metrics. One table with a row per language and columns for spend, conversions, cost per acquisition and revenue tells the whole story in one glance, and the blended figure goes underneath as a total rather than at the top as the headline. The rule worth holding is that no chart in the deck should combine the languages without also showing them apart, because the combined number is the one that hides the problem.