Close-Variant Matching: Potential Impact to Paid Search

Beginning in late September, Google AdWords plans to apply plurals, misspellings and other close-variant matching to all exact and phrase-match keywords. Bing Ads is following suit by testing this feature within the United States, although they will be offering an opt-out for now. It’s important for advertisers to understand how this change impacts the Paid Search landscape and to keep it in mind when comparing year-over-year performance metrics once applied.

In order to gauge the potential impact, Rosetta pulled aggregate data across multiple verticals and found that while close-variant matching did drive incremental conversions, the cost associated was 51% higher.  Additional notable impacts include potential increases in cost per click (CPC), a blending of match types and cannibalization within account structures. The context below outlines close-variant matching, impacts to performance and ways to minimize the impact moving forward.

Close-Variant Overview

Close variations include misspellings, singular and plural forms, acronyms, stemmings (such as floor and flooring), abbreviations and accents. This functionality was introduced in 2012 by Google within campaign settings. Advertisers previously had the ability to opt out of close-variant matching, though Google will remove this option in late September. These variants are now going to be included regardless of whether the keyword specifically lives within an advertiser’s account. Google states that 7% of searches contain misspellings and close-variant matching will ensure relevance and prominence for similar queries without additional work for the advertiser.

An example of a close variant would be an advertiser bidding on the exact match keyword of “baby clothes.” The ad would then show for additional terms that the advertiser may not have planned for. For instance, ads may appear for:

  • babby clothes
  • baby clothing
  • baby cl othes
  • baby cloths

Potential Impact to Paid Search Program

By removing the choice to exclude plurals, misspellings and other close variants, Google and Bing are blurring the lines between keyword match types. “True” search queries are being deemphasized in favor of relative consumer intent, which could have potential impacts to Paid Search program performance. This is similar to a trend that was seen with the introduction of Google Enhanced campaigns in April 2013, deemphasizing siloed device targeting in favor of user context.

Rosetta has found that a query such as “shoes” has markedly different intent and value compared to the query “shoe.” With the lack of a dedicated opt-out feature, these two queries will be viewed as the same, minimizing the control that advertisers have.  Rosetta has historically excluded plurals, misspellings and other close variants as a best practice in order to have greater control over optimizing individual queries, maximizing advertiser return on investment. However, a number of advertisers have preferred to opt-in to this feature, helping to quantify the potential impacts across multiple verticals:

  • Decreased efficiency: While the update will bring an increase in traffic, the lack of complete transparency and removal of an optimization layer will undoubtedly result in an increase in conversion volume though at a more expensive cost per action (CPA). Monitoring the impacts to non-brand terms across multiple clients and verticals, Rosetta has seen a 17% increase in conversions, though at a 51% higher CPA.
  • Increase in brand/non-brand CPCs: On non-brand terms, Rosetta has seen more expensive CPCs among plurals, misspellings and other close variants compared to their counterpart, ranging from 2-10% more expensive. On brand terms, Rosetta has seen much more drastic results, with plurals, misspellings and other variant CPCs costing up to 133% more. Due to the potential brand vs. non-brand ambiguity between singular and plural “brand” terms, the potential CPC increases make them more vulnerable. As keywords match to more phrases, Rosetta expects overall CPCs to increase within brand and non-brand terms alike.
  • Inefficient spending due to budget constraints: For clients who are limited by budgets, the increase in traffic will likely result in additional spending on plurals, misspellings and close variants. This could potentially mean a larger proportion of inefficient spend, resulting in CPA inflation.
  • Blending of match types: Historically, “exact match” has given advertisers complete insight and control into a user’s search query. This shift will result in “exact match” keywords matching to multiple query variations, devaluing the match type and adding an additional layer of complexity. “Phrase match” keywords are subject to the same treatment, which blurs the lines between “phrase” and “broad” match types. While Rosetta’s methodology regarding match-type parity will remain unchanged, this expected change to query mapping will be an additional consideration.
  • Cross-ad-group cannibalization: Close-variant matching will open the door to multiple keywords matching to the same query. These impacts could influence campaign structure moving forward in order to minimize the impact.

How Rosetta Helps

Although advertisers won’t be able to opt out of this matching behavior, there are steps that Rosetta is taking to minimize the potential negative impacts to program performance:

  • Bidding on plurals, misspellings and close variants: As a best practice, Rosetta has included plurals, misspellings and close variants as unique entities in pay-per-click accounts. Google has confirmed that continuing to do so will increase the likelihood that queries will match to the “correct” keyword. Matching will be based on ad rank, which takes into account Max CPCs and Quality Score.
  • Ad group/negative keyword structure: Though positive keyword matching behavior will change, Google has confirmed that negative keyword matching will remain unchanged. Through extensive search-query mining, Rosetta is using negative keyword expansion and account restructuring (where applicable), to funnel search queries toward the “correct” keyword. For example, the keywords “baby clothes” and “baby cloths” can be separated into unique ad groups. Implementing the exact match negative of the other keyword in each ad group will ensure that “baby clothes” won’t match to “baby cloths” and vice versa. Rosetta will determine the proper approach going forward based on the impact of individual keywords/queries.

Once the change is in place, Rosetta will closely monitor search-query reports, matching behaviors and program performance to ensure that negative impacts are minimized.

This piece was authored by Rosetta’s Paid Search Managers Courtney Demko and Patrick Hosking.


Leave a Comment