I was interested in seeing how often Google serves brand ads for non-brand searches. I feared this was happening fairly often, as one of my clients was showing a lift in brand sales and a dip in non-brand sales year-over-year.
To see if this was happening, I pulled search log data for a group of clients. The search log data, seen through our redirect logs, shows that this is in fact happening, but only about 0.5 - 1% of the time.
Given this, the client’s brand spike and non-brand dip are only slightly related to the search log findings, and instead can be attributed to other variables; including off-line branding efforts, changes in site conversions, more aggressive non-brand competition year-over-year, and other factors.
I pulled all the searched keywords and RKG ads served/clicked on Google for a handful of clients’ over a 90 day window. (This data is also very helpful to show if we should expand the keyword list further and/or add negatives where client ads should not be served.)
I looked at instances where a brand ad was served and clicked when a non-brand keyword was searched. To do this, I filtered the RKG ad column (the Google ad that was served and clicked on) by the brand regular expression - which is how we define a keyword as a brand keyword. Then, I filtered the search log column to only include non-brand keywords (exclude the brand keywords.)
(Sometimes we are unable to see the search string, so this portion of traffic is unknown and so not included in the analysis.)
Depending on the client, the data shows that this is happening for less than 1% of the total Google traffic.
Although our keyword lists are quite extensive, we do not have every non-brand searched permutation. In these cases, Google is taking liberty to show a brand ad, rather than no ad at all or another non-brand keyword that is more relevant and targeted. This is happening because Google doesn't classify terms as brand or non-brand on its end. Instead, Google is most interested in serving the highest quality ad available during each auction. Since brand ads show a higher CTR, Quality Score, and bid level in Google's system, these are more eligible for auctions where we do not have an active keyword matching the user's search query exactly.
But, we’d argue that in the cases where there is an active keyword in the account that matches the search string, Google should serve this keyword, as it is the most relevant and targeted ad.
Why Do We Care?
Because this is only happening a small fraction of the time, I am less concerned. But, it is important to point out, that for those non-brand searches, the user is being shown a brand ad, which behave differently than non-brand/incremental ads. Users do not have the same intention when searching for non-brand keywords, and so the closing ratios are lower on non-brand ads. As such, showing brand ads when the user searches for a non-brand keyword may deflate the closing ratio in the brand category. (This is only slight, given the data I pulled.)
Also, the copy and landing page are not always relevant and targeted to the non-brand search. This too can hurt conversion in the brand category.
The new Broad Match Modifier option may help keep brand ads from serving on non-brand searches. Another option, though tedious, is to add negatives to the brand adgroups that are firing ads when users are searching for non-brand keywords.