Delayed Growth of Close Variants Following Google Changes Now Becoming Clear

Last June I wrote about what the early data was telling us following Google’s March 2017 adjustment to the definition of close variants. This change made the addition, subtraction, and change of function words acceptable adjustments for a query to be deemed an exact match close variant, and also allowed for the words in a keyword to be reordered and still count as an exact match close variant.

Changes to Close Variants

At the time, it appeared that there were no meaningful shifts in the types of match types that were driving traffic, and that Google’s adjustment to the definition of a close variant might not be of much importance at all.

However, beginning in Q3 we began to see an increase in the amount of traffic coming from close variants. Let’s take a look at how much traffic is now being attributed to exact match close variants, as well as how that traffic is performing relative to pure exact match.

Note on Method: In determining close variant share and relevant performance, I use Google’s match type designation in the search term report. This does not reflect the match type of the keyword but rather the relationship of the query to the keyword. This isn’t the only way to look at such match type comparisons, and each method of comparison will be impacted to some extent by advertiser strategy, such as the deployment of negative keywords and the extent of keyword coverage. Still, it gives some idea of what the close variant update has done to traffic mix.

I also focus on non-brand traffic, as close variants account for only a very small share of traffic for brand keywords and perform much more similarly to pure exact matches than is the case with non-brand.

Close Variants Now Account for 20% of Desktop Exact Match Traffic

As you can see from the chart below, the share of exact match non-brand text ad traffic coming from close variant queries went up meaningfully in Q3 across all three device types and stayed at heightened levels through Q4.

Close Variant Exact Match Traffic Share

For desktop, this surge resulted in 20% of all exact match traffic coming from close variants in Q4, up from just 12% in Q2 2017. Phones, which have long seen lower share than the other two device types, saw 14% of exact match traffic coming from close variants in Q4 2017, compared to 11% in Q2 2017.

This is important to advertisers because close variants typically convert at a lower rate than true exact match traffic. While an influx of newly minted close variant queries might have caused relative conversion rate to shift, we find that close variant conversion rate relative to pure exact match has remained about the same since the uptick in traffic.

Close Variant vs Pure Exact Conversion Rate

Thus, with such an influx of traffic, we might expect the conversion rate of overall exact match (including close variants) to suffer. However, we see quite the opposite.

Non-Brand Exact Match Conversion Rate Ramping Up

Looking at aggregate conversion rate of exact and exact (close variant) matches for non-brand text ads, we find the median advertiser saw a meaningful increase in the back half of 2017 for both phones and desktop.

Exact Match Conversion Rate

This follows an increase observed in overall non-brand sales per click across Google Shopping and text ads, which we reported on in our latest Digital Marketing Report.

Non-Brand Revenue Per Click Growth

The exact cause of these increases is still a bit of a mystery. In the big scheme of things we do expect conversion rate to naturally increase over time with steadily evolving methods of targeting, better websites, better devices, and searchers’ growing comfort with converting online. However, the jump observed in the back half of 2017 represents a steep acceleration that would seem too much of a shift to be attributed to these slowly evolving variables.

One potential explanation for this increase is that Google’s May change to Ad Rank did effectively place better ads in front of users, resulting in a higher conversion rate. The change reportedly adjusted Ad Rank thresholds to account for the meaning of the query, and also meant that bids might be weighted more heavily in Ad Rank calculations depending on the meaning of the query.

This update appeared to result in a significant increase in first page and top of page minimum bid estimates for non-brand keywords which continued through the end of the year.

Google Minimum Bid Estimates

This has coincided with a steep acceleration in Y/Y CPC growth for non-brand traffic.

Google Non-Brand Growth

Thus, it’s possible Google’s May update increased both the price that advertisers pay for traffic, as well as the value of that traffic to advertisers.

Back to the topic of close variant traffic, we should have expected the aggregate conversion rate of exact and exact close variant traffic to decline with the influx of additional close variant clicks. However, with the overall rising tide of non-brand conversion rate in the back half of 2017, no such decline was observed.

Additionally, the expansion to the definition of close variants might have helped click growth along by including ads in auctions that might not have otherwise featured ads at all or as many ad units. However, we saw a steep deceleration in Y/Y click growth in the back half of 2017, so any incremental traffic that was the product of the change wasn’t enough to prop up click growth.

Conclusion

While the early data suggested that the March update to the definition of close variants did little to shift traffic to the exact (close variant) match type, Google did say that this update would be rolled out ‘over the coming months,’ and the back half of 2017 told a different story. Thus, this is a great example of how some updates take time to really show up in the data.

Regardless, the increase in close variant traffic probably should have caused an overall decline in non-brand exact match conversion rate, and seeing increases in this metric as close variants ramped up is certainly a positive development.

However, it is still the case that advertisers must continue to assess the performance of the close variant traffic that is matching to each keyword, since the lower conversion rate of close variants can still reduce the overall value of an exact match keyword. This might result in less competitive bids than would be the case if the keyword was only receiving true exact match traffic, potentially resulting in worse position and fewer clicks.

Brands can combat this by adding negatives for those close variants which are harming performance. Google may have loosened the amount of control that match types place on query matching, but marketers still have the ability to prevent some matches from driving ad traffic.

Theoretically, brands should be able to add those close variants which they do want to target (just not with other keywords through close variant matching) as keywords on exact match so that Google sends all traffic from these queries directly to the correct keyword that matches the query perfectly. However, it’s long been the case that examples exist of Google choosing a less closely-matched keyword over one that would match it exactly and which is placed on exact match, in contrast to Google’s stated policy of giving exact match keywords ‘preference’ whenever possible.

Therefore, to truly rule out the possibility of another keyword garnering traffic for a specific close variant launched as its own keyword, a corresponding negative keyword will need to be added for other keywords which might ‘steal’ the traffic. The expanding definition of exact match close variants has meant that even more keywords need to have such negatives added in order to usher traffic to the correct version, and thus grown the importance of mining query reports for potential optimizations.

Join the Discussion