Google killed manual language targeting. Here's what that means for your CPA.

For years, running a multilingual campaign meant setting a language target and trusting the platform to route locally-written copy to the right audience. That target was never a quality check, but it was at least a visible setting: get it wrong, and something in your account clearly looked wrong.

That setting is gone. Google Ads has phased out manual campaign-level language targeting for Search entirely. In its place is fully automated matching: the same AI infrastructure behind AI Max and Advantage+ now decides which ad copy reaches which user, blending multilingual delivery into ordinary ad serving.

For an advertiser, this is not a convenience. It's a transfer of judgment. The platform used to need a language target to route ads at all; if that target didn't match the copy, something visibly broke. Now the routing model matches copy to users algorithmically, on signals like predicted relevance, whether or not the copy is actually correct for the person receiving it. A literal translation doesn't trip a targeting mismatch anymore. It just gets served with full algorithmic confidence to someone it was never really written for.

Three things follow from this shift:

1. Manual QA is no longer optional. If the platform isn't catching a bad translation, someone has to, before the ad goes live, not after the CPA report comes in.
2. Match rate isn't enough. A string can be 100% covered by translation memory and still fail. If the underlying phrase is an idiom, a claim, or culturally specific, a "perfect match" score means nothing about whether it actually works.
3. Speed still matters. The fix can't be a five-day review cycle. It has to run inside the same tempo as the platform's own automation, which is now continuous.

That's the actual argument for a risk-routing engine, not as a translation upgrade, but as the check the platform itself no longer performs.