Match type on a negative keyword

Negative keywords enter an account from three places: added from the search terms report as a quick fix, set at the campaign or ad group level while building the account, or dropped into a shared negative list applied across several campaigns. Whichever layer it sits in, a negative carries a match type the same way a positive keyword does, and that choice travels with it.

The blast radius differs by where it sits too. Negatives set at the campaign level blocked a converting query 4.63% of the time, ad group level negatives did so 2.05% of the time and shared list negatives 1.89% of the time, each measured across at least 5 accounts. Layer matters, but match type matters more, which is why an audit has to open every list, not just the shared ones.

A negative's match type is not the same instrument as a positive keyword's. Google is explicit that negative keyword match types work differently from their positive counterparts and do not match to variants, and that negative keywords can be exact match, phrase match or broad match.

What the measurement shows is that the three types do not carry equal risk, and not in a straight line either. Broad match negatives blocked a converting query at 11.96%, far above exact match at 1.75% and phrase match at 0.61%. Phrase sits below exact rather than between exact and broad, so broad is an outlier rather than one end of a smooth gradient. Broad negatives were only 5,235 of the 75,695 measured, a small slice of the list carrying a disproportionate share of the blocking.

A negative that matches a query stands between the account and that query, and where the query had converted before the negative was added, that demand stops arriving. It stays that way until somebody audits the list. In this measurement, 11.96% of broad match negative keywords blocked at least one historically converting query, against 1.75% of exact match negatives, across 7 accounts and 75,695 live negatives.

That is not a one-off click cost. It is a standing exclusion that outlasts whoever wrote it, and counting the number of negatives in an account tells you nothing about how much of it is exposed this way. A list built on habit rather than a deliberate check of match type is a list with a hidden floor under its revenue.

Google's own interface defaults to the safer option: negative keywords added to a Search campaign from the search terms report are added as negative exact match, and changing the match type means using the appropriate symbols. Every broad match negative in an account is therefore one somebody widened, deliberately or without noticing, rather than something the platform chose.

For a founder comparing agencies, this is a concrete thing to ask about: does the agency audit negative match type, or only negative keyword volume. For an in-house practitioner it is a recurring task rather than a one-time clean-up, because every new broad negative added under time pressure joins the same population measured here.

This is not a claim that broad match negatives are always the wrong choice. Plenty exist precisely to block a wide category of irrelevant traffic on purpose, which is what they are for. It is also not a measurement of money lost: it counts negatives that intersected a query with a conversion history, not the revenue that query would have produced from then on.

It is not a statement about how Google's auction or matching engine works beyond what Google itself publishes: that negative keywords can be exact, phrase or broad match, and that the right one is whichever does not prevent your ads from showing to people searching for your keywords. The number here is a prompt to check match type, not a verdict on any single negative.