Scott Shannon’s “Keyword (Not Provided) & The End of SEO Data” makes a great case for SEO adaptability. Shannon posits that just like when someone loses a sense and the other senses get stronger, the same is true when we lose data. On average, between 40 to 60 percent of websites are missing keyword-level data. Keyword-level data has become so ambiguous that even trying to proxy back in the missing data using weighted allocation is becoming pretty useless, especially with so much going into the unknown bucket. One would think that this conundrum makes the job of being an SEO manager more difficult, but in reality resourceful SEO managers have found their workarounds as needed. With a little understanding of how their keywords impact marketing strategies, SEO managers can begin the process of effectively building higher-level strategies, thus making the jump from keyword level data to page-level data effortlessly.
All in all, I believe “keyword not provided” to be more of a testing and measuring issue only when attempting to look at the performance of specific isolated keywords. I know how challenging it can be to look at keyword-level metrics in isolation with a high degree of confidence. However, I have found that using historical performance and paid search data can help normalize the data if there are strong enough correlations. That said, there are many other angles for looking at the data to compensate for the missing keyword data depending on your situation. Tools like Adobe Site Catalyst and Discover are also highly relevant resources for acquiring SEO data. On the plus side, “keyword not provided” has been less of an issue with co-optimizing with paid search tests where there are workarounds using control and treatment groups.
I have come across quite a few websites with great “keyword not provided” workarounds that I have found beneficial. Be sure to check out The RKG Blog, Occam’s Razor, Econsultancy, and Search Engine Watch for a few pointers.
In closing, although the keyword data loss definitely has the capacity for strengthening other SEO data strategies, we can and should still learn from all the SEO data that we have.