Using Google Analytics to Analyze Not Provided Keyword Data

On October 18, 2011, Google announced that search query data for logged in users would no longer be available.  As marketers and analysts, we know that every bit of data is precious to us.  Every piece of data can have a huge impact on the major decisions we make as businesses.  So how much impact has this privacy change had on our data?  Google claimed it would only impact a small amount of data, in the 2-3% range.  Is that holding up? Let’s take a look at the impact on one of our clients:

Date Range: Oct. 18-Feb. 10. 

Total Organic Visits: 730,403

Total Not Provided: 60,138

Not Provided is responsible for about 12% of organic visits to this site, a large enough chunk to make it worth further analysis.  So, let’s try and find out a bit more about Not Provided.  First, I set up a Custom Report with all of the data I thought would give me an idea of the type of user Not Provided is.  Here’s what my report looks like.

Within my custom report I looked at only the organic traffic.  Now, I want to divide my traffic by brand keywords, non-brand keywords, and Not Provided to see if there are any commonalities.  To do this I used three different Advanced Segments.  For the first segment I included only brand terms, for the second I excluded brand terms as well as Not Provided and for the third I only included Not Provided.  I decided to look at the data starting on Nov. 1 because that is when Not Provided leveled out and the traffic became more consistent.

(Blue = Brand, Orange = Not Provided, Green = Non-Brand)

The not provided user most closely resembles the brand users.  However, remember that correlation is not causation.  Let’s take this to the next level and see what information we can glean from the landing pages.  To do this, I looked into my standard Organic Keywords report, drilled down to Not Provided and set up my secondary dimension as Landing Page.  My top five landing pages for Not Provided consist of my home page, special offers page, and major product pages.  These five landing pages account for 8,277 visits.  In other words, about 42% of Not Provided users are landing on five very branded pages. 

Now I would suggest drilling into each of the top Not Provided landing pages and then looking at what other keywords are landing on each of those pages.  This will give me an idea of what kind of keywords are bringing up that landing page and then I can guess that my Not Provided keywords are probably variations of those keywords.  After looking into the keywords that landed on my top five landing pages for Not Provided, I can see that almost all of them are brand related keywords.  In this case, my landing pages report supports my theory that a big portion of the Not Provided keywords consists of branded traffic.  From looking at the sheer quantity of other landing pages Not Provided has, I could make a pretty good guess that a lot of those are long tail keywords. 

When I saw Not Provided as my number one keyword, I got worried about all the information I couldn’t see.  However, after some analysis I can see that Not Provided doesn’t consist of a few magical keywords that I am suddenly missing out on.  Instead, I would say about half of them are brand keywords and the other half are probably long tail keywords that I wouldn’t normally give much thought to if they weren’t all clumped together in the Not Provided bucket.  

What tools are you using to gain insight into the Not Provided keywords?