Google’s latest, machine learning driven, changes to which queries will be considered as ‘exact match’ to a keyword takes the intent of a search into consideration and could change the way you research and assign keywords.
As far back as 2014, Google began making changes to its ‘exact match’ Ads function – first including plurals and, later, misspellings in ‘exact match’ results. Their latest update, however, uses machine learning to determine the intent of a query, and return ‘exact match adverts even if the query contains some slightly different terms to your keyword. How does this work in practice? Google have released the table below to demonstrate for the keyword ‘Yosemite camping’.
Google’s reasoning behind this is that 15% of daily searches are new, so keywords relevant to your Ads may not have been discovered through previous keyword research. The new exact match function will allow you to display Google Ads for relevant queries without having to build exhaustive lists. According to Google, advertisers using mostly exact match keywords saw a 3% increase in exact match clicks during the pilot study of the new feature.
Even since this update went live in October, Google still prefers keywords identical to the search query where possible, but all in all you should expect to begin matching with new queries now this feature is live. Whilst Google expects this feature to be beneficial to most of its advertisers, its worth remembering that no machine learning system is (yet) perfect, and that you may begin matching with some queries you don’t want to display ads on, so its worthwhile keeping a close eye on new queries and updating your negative keyword lists regularly.
Phrase, Broad Match and Broad Modified results have not been affected by this update, so if you are wishing to display Ads for a very narrow set of queries, it may now be better to use Phrase match, as all words in the phrase must still appear in the query.
Have you seen a change in your Google Ads results since October? Tell us in the comments below.