In October Google announced that they would be encrypting searches if the user was logged into Google when searching on Google.com. What this means for your website is that in Analytics data you will see in the organic keyword list (not provided) as a keyword. While the percentage of searchers that are signed into Google might be quite low it can affect your reporting in a number of ways.
I like to report on the split between branded keywords and non-branded keywords to show how SEO efforts are increasing traffic to the website, month on month or year on year. However the data is now skewed because of secure search and the (not-provided) keyword. In Analytics you would usually exclude keywords that contain the brand name and work out what percentage of the keywords is branded and what percentage of keywords are non-branded. As the keyword (not-provided) does not contain the brand name it will be included in non-branded keywords which means non-branded traffic will therefore be inflated. So how do we get a more accurate number?
Method
This method is based on the assumption that signed-in searchers and other searchers behave in the same way.
Example:
Total Organic Search Traffic = 20,000 visits
Total Branded Keyword Traffic (excluding (not provided)) = 8,000 visits
Total Non-Branded Keyword Traffic (including (not-provided)) = 12,000 visits
(Not-Provided) Traffic = 2,000 visits
Step 1: Finding Total Non-Branded Traffic Excluding (not-provided) keyword.
Total Branded Keyword Traffic (including (not-provided)) – (Not-Provided) Traffic = Total Non-Branded Keyword Traffic (excluding (not-provided)).
12,000 visits – 2,000 visits = 10,000 visits
Step 2: Working out the percentage split between Branded and Non-Branded Traffic
Now you need to work out the percentage split between Branded and Non-Branded Traffic.
So Total Branded Keyword Traffic (excluding (not provided)) + Total Non-Branded Keyword Traffic (excluding (not-provided)) = Total Traffic (excluding (not-provided)
8,000 visits + 10,000 visits = 18,000 visits.
Branded Keyword Percentage = (8,000/18000) * 100 = 44%
Non-Branded Keyword Percentage = (10,000/18,000) *100 = 56%.
We now know that typically this website gets 44% Branded traffic and 56% Non-Branded traffic.
Step 3: Working out the Branded and Non-Branded Split of the (not-provided) Keywords
We can now take the (not-provided) keywords and split this number according to the split we have just calculated so;
(not-provided) branded traffic = 2,000 * 44% = 880 visits
(not-provided) non-branded traffic = 2,000 * 56% = 1120 visit.
Step 4: Work out actual Branded and Non-Branded Traffic Totals
Now we can work out a more accurate number of visits for branded and non-branded keywords.
Total Branded Keyword Traffic = 8,000 visits + 880 visits = 8880 visits.
Total Non-Branded Traffic = 10,000 + 1200 = 11200 visits.
We can now use these numbers to compare to last month or last year more accurately. This is obviously a quite simple solution to a small problem but as you dig deeper into your analytics data much bigger problems are bound to come up. How will we tackle these problems as natural search marketers? Will be become less keyword focussed and focus more on the keyword set as a whole? I don’t know the answers to these questions but time will tell and because Google is the dominant search force it is up to us to adapt to the changes they make, whether this is an algorithm or keyword data change.
Have you found any challenges in reporting since Google started protecting its users’ searches? Let us know.
Adrian Jennings
Search Engineer