In my column last week, I explored what impact Google’s controversial preclusion of some search keyword queries has had on our industry. I also confirmed the current percentage of affected terms (20.5%), noting the more than 100% growth in just a year’s time (Google originally stated less than 10% of search queries would be affected).
After that piece was published, I continued to think about (not provided) and what it really means for webmasters and SEOs. Conversations with friends and colleagues only fueled my curiosity to explore further. What really strikes me as interesting, and disappointing, about (not provided) is how it will limit our insight into search personalization.
If you subscribe to the idea that search results will only continue to become more personalized over time (through smarter algorithms, additional social intelligence, and incremental knowledge of user behaviors), it’s logical to assume that authenticated Google search users will see the most personalized results sets. What (not provided) eliminates, then, is the potential to explore the nuances of personalized search, and how personalization impacts subsequent on-site behaviors. All we have now is keyword referral data from vanilla, default search results.
I’ve long been obsessed too with the idea of identifying “rank at the time of the click,” and replacing standard SEO rank reports with these more meaningful data points. Imagine what could have been: rank-at-the –time-of-the-click details for long-tail, personalized search results. Keyword and content optimization recommendations would have jumped off the pages of analytics reports! Identifying instances where engagement and conversion were strong for terms that only seldom appeared across the results pages would have been easy work. The visitor segmentation options would have been dizzying.
And no, I don’t think paid search gives us the same type of insights for numerous reasons, and Bing doesn’t (yet) have the market share to come close what we could have gotten from Google, especially across B2B industry verticals.
So what the (not provided) are we supposed to do? As a friend pointed out to me on Facebook, the most difficult part of life in the post- (not provided) world is explaining the data gap to clients. Defending the value delivered through SEO services is harder when you can’t completely identify the organic search terms sending traffic.
Fortunately, we’re not without recourse.
Numerous how-to articles have appeared over the past year plus, detailing how best to overcome the loss in data and still make smart optimization decisions. This isn’t a complete listing, but it represents my favorite approaches.
1) Get everyone to switch to Bing. C’mon, who’s with me? I can hear Frank the Tank from the movie “Old School” yelling, “We’re going Binging!”
If that doesn’t work…
2) Approximate brand versus non-brand impact. A decent, directional approach to take. Dave Davies on Search Engine Watch offered up his thoughts in a February column, “How to Understand Your Google ‘Not Provided’ Traffic.” This approach entails investigating on-site engagement patterns from known keyword referrals, then using that intelligence to approximate the terms that are missing. In his example, he uses this method to delineate between brand- versus non-brand terms, but notes that specific keyword themes can be used as well.
3) Cluster keywords by landing page. On the heels of Google’s announcement last year, AJ Kohn wrote the excellent piece, “Not Provided Keyword Not A Problem.” His advice then stands today as one of the best approaches to addressing (not provided). His recommendation is to analyze organic keyword referrals by landing pages. Keyword (not provided) will be included, but the mix of other terms will provide directional evidence of the terms that are not being passed through to analytics.
4) Insert page names in place of custom keywords. Just two weeks ago, Carrie Hill on Search Engine Land described what I believe is the best approach to handling (not provided), with “How To Turn (Not Provided) Into Useful, Actionable Data.” This method calls for custom filters within Google Analytics to capture and report the landing page URL as the keyword. So rather than seeing (not provided) in analytics reports, this would instead produce entries such as “np - /page-specific-topic-and-keyword-target/.”
Note that this type of filter can also be applied to populate the keyword field with the page’s title in lieu of URL, assuming that title is meaningful. Also, methods 3 and 4 both assume a logical URL taxonomy and the presence of basic SEO keyword targeting.
I think it’s safe to conclude that griping about missing keywords isn’t going to get us anywhere. That ship has sailed. What is required of SEOs now and in the future is planning for what the (not provided) to do about keyword (not provided).