In late-September last year, Google rolled out mandatory close variant matching
(CVM) to all phrase and exact match keywords. The change forced phrase and exact match keywords into auctions for queries that were deemed to be closely related to each keyword, such as plurals and misspellings. Shortly after the change Merkle | RKG reported slight shifts
in the share of traffic coming from each match type across RKG’s client base, the vast majority of which had refrained from opting into close variants prior to the mandatory change. Now a few months in, let’s take a look at how these close variants are performing and impacting the keywords they’re being matched to.
Conversion Rate Difference Significant for Exact Match
Google’s decision to cease passing users’ search queries
in the referrer string for paid search clicks in April of last year eliminated advertisers’ ability to analyze the impact of CVM using third party conversion data. However, including the match type column in Google’s search term report allows advertisers to view performance of true exact and phrase match traffic versus CVM traffic. If Google conversion tracking has been set up, advertisers will be able to compare conversion rates as well. Across a sample of RKG clients, we see that close variant exact match search terms see a conversion rate of just 48% that of true exact match search terms, while close variant phrase match conversion rate was 86% that of true phrase match. These differences in conversion rate exist for both branded and non-brand keywords.
That’s a huge difference in conversion rate for exact match keywords. However, when we look at the share of exact and phrase match traffic coming from close variants, exact match keywords see a much smaller share of their traffic (4%) coming from close variants than phrase match keywords (26%).
This is likely because RKG advertisers have not had to limit close variant phrase match with the use of negatives as much as exact match due to the closer conversion rates of CVM and true phrase match, such that close variant traffic hasn't been as much of a detriment to performance for phrase match. Google also has more flexibility with phrase match keywords since words can be tacked on before or after the keyword, allowing phrase match keywords to match to a broader array of close variants than with exact match keywords. Looking at the share of traffic that is coming from each match type over time, we see that not much has changed since the initial bump in phrase and exact match traffic share in late-September.
We have also not had to deploy negatives any more aggressively than was necessary when the change occurred. This indicates that there haven’t been major changes to what types of variants Google considers ‘close’ or to the frequency with which Google deploys close variant matching. As the share of exact match traffic coming from CVM has remained small, exact match sales per click relative to broad match stayed relatively the same year-over-year, as did phrase match sales per click, and each actually saw slight bumps in January.
Overall, Google doesn't appear to be changing the rate at which it is serving close variants, and the current levels of traffic aren't having a huge impact on overall sales per click for each match type. However, the limited impact for RKG advertisers has been strongly influenced by campaign management aimed at minimizing the impact of CVM on performance, and hopefully other advertisers out there are doing the same. Our message remains that marketers should not feel confident that close variant matching is fine in all cases. The impact stands to be unique for each program, campaign and keyword, and managers need to stay vigilant to prevent cases in which CVM is harming account performance, especially if Google ever moves to broaden its definition of ‘close variant.’