5 Ways to Work Around Not Provided Keywords
UPDATE: January 13, 2014. On December 31, 2013 Google updated GWMT search query data to be “less rounded”. If you visit GWMT you’ll see a line marking where the change took place.
UPDATE: September 30, 2013. Last week we found out that Google is extending SSL protection to users that are not signed in. This means that for ALL organic search traffic that Google sends to your site, the keywords will be removed and replaced with “not provided”. At this point in time, Bing is continuing to provide webmaster keywords in the referrer string.
Starting in October 2011, keywords used in secure searches are now hidden from webmasters by Google. This means that in Google Analytics (or in any other analytics tool), keywords from these searches show up as not provided. A secure search could mean the user is logged into a Google account, is using Firefox 14 or is just explicitly using secure search. A lot has been written about this controversial move, which was done ostensibly to protect privacy but did not extend to Adwords.
Although Google initially stated that the change would affect 10% of queries, there are many sites where the percentage is much higher. In WebEnso’s case, keywords for 62% of searches fall under not provided, and the percentage is still climbing. 62% is a huge percentage. This means that more than two out of three keywords used to find my site is hidden from me. So what to do?
Five Ways to Work Around the Not Provided Black Hole
Search queries in Google Webmaster Tools
Under the Traffic menu in GWMT you will find search queries. This will show you the top 1000 search terms as well as other useful information such as average rank and click through rate. Keep in mind that the GWMT data only goes back 60 days, so if you want to track keywords trends for your site over time plan on regularly exporting the data. In GWMT not only can I get a better idea of what the top search terms actually are for my site, but how am I doing on click through. For example, even accounting for the lower position, the low CTR might be telling me that my article on WordPress robots.txt is not exactly what searchers are looking for or my snippet needs work.
Bucket your keywords
We may have lost the ability to fully analyze keywords individually, but we can still mine Google Analytics for interesting data if we take a more macro view of the data. Not all keywords are created equally and users will be behave differently if they are searching on your product or brand terms versus a generic search. You can use Google Analytic’s custom segments to bucket your keywords, for example if you create a segment that exclude all brand keywords, does that traffic behave differently, eg. is the bounce rate higher? Most branded searches convert at a higher rate than generic searches.
For the not provided bucket, although the keywords vary, we do know that much of the traffic is from people who have a Google account and are logged in. My theory is that these are people likely to be in the technology industry in some way. Even though I can’t see the keywords they are searching on, I can see their top landing pages, and user behavior signals such as time on site and bounce rate. To create a custom segment click on “Advanced Segments” and then on the “Create a Custom Segment” button and then fill in the conditions (see below for creating a custom segment for not provided keywords).
Doing this exercise I found a top landing page that I wouldn’t have thought was ranking based on the limited keyword data available to me in Google Analytics. For a much deeper dive into analyzing the not provided traffic see this Secure Search Analysis post from Avinash Kaushik
Queries in Google Analytics
Your keywords are also available to you in GA’s Queries (Traffic Sources -> Search Engine Optimization -> Queries). It is two calendar days out of date and you will have limited ability to analyze the data (ie. you can’t add landing pages as a secondary dimension for example). But who would have known that my post on the Internet Blackout to Protest the SOPA legislation would be getting traffic for “sad koala”?
You will need to link Google Webmaster Tools to your Google Analytics account to see this data. See “Configuring SEO Data in Analytics”.
Keyword analysis per landing page
Do landing page keyword analysis. First navigate to Content -> Site Content -> Landing Pages, click on the page you are interested in and then add “Keywords” as a secondary dimension. This will give you the search term used to find that page and theory is that you focus on the keywords that are listed. Note that we are assuming the keywords hidden from you as not provided are not going to be that much different than the ones you can see. This technique is very useful for sites where not provided is not such a big percentage of their keywords – as you can see below, not provided makes up such a huge portion of keywords for my site that I don’t feel that I have the complete picture.
Run a PPC campaign
The hiding of keywords from secure searches does not extend to data collected for Adwords PPC campaigns. This fact irritates many SEOs, but putting our feelings aside about having to pay to get more complete data – it is another way you can get keyword data. So if you have the budget and want to understand the keyword performance in your market quickly, consider creating a PPC campaign. Get access to your keywords with Matched Queries (under Adwords menu). Note this will only show the keywords you bid on, but it could be another source to getting a more complete pictures of your keywords.
If the above approaches are not enough for you, you could put on your Ninja hat and install a script that overwrites the not provided keyword in the referrer string with the inferred keyword. Approach with caution.
BTW, if you are annoyed by Google hiding keyword data in this way, you can sign the notprovided petitions.
Latest posts by Kathy Alice (see all)
- Checklist for Choosing a SEO Friendly Content Management System - January 20, 2015
- Can your site get a Bing Penalty? - December 30, 2014
- Does User Experience Matter for SEO? - November 18, 2014