PMax can start bidding in languages you never targeted

Performance Max is built to expand reach automatically, matching queries loosely against your assets and audience signals. That is a strength when it finds genuinely adjacent demand, and a real cost when it drifts into search categories the account was never set up to serve.

The concrete, checkable version

One concrete, checkable version of this: search categories that show up in scripts the account never targeted -- Arabic, Cyrillic, Chinese, Japanese characters appearing in a campaign built for Latin-script markets. Google does not report cost per search category inside PMax, so this specific leak never shows up as a spend line -- the only signals available are impressions and clicks per category, which is exactly why it goes unnoticed: there is no euro figure to trigger an alert on, only volume drifting somewhere it should not be.

A worked example

Illustrative case: a PMax campaign built for a Dutch and German market has been running for four months with no reported issues. A search-category review turns up several thousand impressions and a modest but nonzero number of clicks in categories written in Cyrillic script -- no euro figure attached to flag it, since Google reports impressions and clicks at that level but not spend. Left unchecked, that drift does not show up as an efficiency problem in the account, it shows up as reach quietly spent on a market the account was never built to serve or convert.

The fix

The fix is not to distrust PMax's expansion, since reaching adjacent demand is the point of the campaign type. It is to check, on a cadence Google's own dashboard never prompts, what the expansion actually reached -- and to treat a search category outside the account's targeted languages as a language-and-market question, not a performance number to average away.

Why it is easy to dismiss as harmless

The instinct, once this turns up, is to shrug it off as a rounding error -- a few thousand impressions against a much larger campaign total. That instinct is exactly why it survives unnoticed for months: each individual check finds a small number, and nobody adds the small numbers up across a full quarter to see what they represent as a share of reach that was never going to serve the account's actual market.