Demystifying Google AdWords Keyword Variants

Within the settings of every Google AdWords Search Network campaign is a feature designed to make advertisers’ lives simpler. But since its inception, it has caused more angst than ease. This feature allows for variant terms to show based on exact and phrase match keywords.

Exact and phrase match variants.

The idea is that advertisers don’t have to manually bid on plurals, misspellings, and other close variants of the root keywords. For example, if you are bidding on the term “blue widget” your ad could be triggered by these queries:

• Blue widgets (plural);
• Blue widdgets (misspelling);
• Blu widget (close variant).

In theory, the feature is a helpful concept. Rather than adding hundreds of plurals and misspellings, you are able to bid on one term that should capture all of these variants. Not only are you saving time, but your ads are also receiving more exposure. Unfortunately, with this option, advertisers do give up some control by letting Google define what a close variant is.

As with other features that put management in Google’s hands, such as Dynamic Search Ads and automated bidding strategies, advertisers are hesitant to relinquish control.

To determine if the close variants option is worthwhile, I reviewed data from five accounts during the time period of Jan. 1, 2013 to Jan. 31, 2014. Before diving into the details, here are a few notes about the study.

• The close variants option was turned on in each individual campaign during the time period.
• Campaign and ad group level negative keywords were added on a weekly basis.
• If a plural variant did not have the same meaning as the singular version (or vice versa) that term was added as a negative. For example, if only one “patriots hooded sweatshirt” was in stock then “patriots hooded sweatshirts” (note the plural “sweatshirts”) was added as a negative.
• This data does not include branded campaigns.

Percentage of Variants

The table below looks at the variant percentage of clicks, costs, conversions, and revenue in comparison to the account. For example, 9.51 percent of Client A’s total clicks came from variant matches.

Variant data.

The common thread across all five clients is that on average, less than 10 percent of overall clicks, costs, conversions, and revenue were from variant terms. In other words, about 10 percent of the costs and traffic went to terms we would not have had exposure for without the variant option.

Cost Per Conversion

Cost per conversion data.

Across all five clients, variant cost per conversion was higher than non-variant. With Clients C and D, variant cost per conversion was only $0.45 and $0.15 higher than non-variant, respectively. Clients A and E saw a larger gap with variant cost per conversions being approximately $1 higher. Client B saw the largest gap with cost per conversion being more than $5 higher.

Even though all clients had larger variant cost per conversions, only one was more than $1.63. In fact, variant costs per conversions were still in line with overall cost per conversion goals.

Return on Ad Spend

Return on ad spend data.

For every client, variant ROAS was lower than non-variant. Just like cost per conversion, Client B saw the biggest difference between non-variant and variant ROAS. It’s important to keep in mind that, on average, variant revenue only accounted for 8.11 percent of total revenue.

Average Order Value

Average order value data

For Clients A, D, and E, variant average order value was actually higher than non-variant. Additionally, Client C’s variant average order value was only $0.11 lower. Only with Client B was variant average order value significantly lower (which follows the trends of cost per conversion and ROAS).

Conclusion

Based upon the data, in most pay-per-click campaigns advertisers should include phrase and exact match close variants. With four out of the five clients, variant metrics were in line (and acceptable) with non-variant. If this option had been turned off, these clients would have lost about 8 percent of their total conversions and revenue over the last 13 months. In general, the searcher’s intent for variant terms was in line with the non-variant option.

The data also illustrates how important it is to actively optimize campaigns. Even with proper management, variant metrics can show significant changes compared to non-variant, as evidenced with Client B. That’s why it’s important to continually add negative keywords and even turn off the variant option in individual campaigns.