Why am I seeing (not provided) in Google Analytics? Updated 21st October 2011
If you like many people over the past few days have started wondering why this is starting to show up in your Google Analytics reports then read on:
This is due to a change that Google made on 18th October 2011 to move all signed in Google users to SSL encryption. This means that all search query data from signed in users now goes over https:// and users’ search query data is now hidden from the beady eyes of SEOs everywhere.
Now, when a signed-in user visits your site from an organic Google SSL search, all web analytics software should recognise the visit as a Google “organic” search but the keyword phrase the user typed will no longer being shown – it will instead be shown as “(not provided)”. I presume they’ll localise this for international customers?
However, a web site accessed through organic search results by non signed-in users on http://www.google.com (non-SSL) will still be able to see in their analytics that both that the user came from google.com and their search query.
On the Google Analytics blog they explain the change and add this lovely one liner…
“Keep in mind that the change will affect only a minority of your traffic.”
Read the full post on the Google Analytics blog.
This change has sparked plenty of debate from privacy advocates, industry commentators and SEOs.
Google’s official blog post went up on the 18th October and we started seeing (not provided) data the same day – since then many more seasoned SEOs and industry commentators than me have voiced their opinion about the change:
Danny Sullivan on Search Engine Land
Pete Young – on Holistic Search
Shaun Anderson on Hobo Web
The general consensus seems to be that Google is behaving in a manner which is inconsistent – on the one hand they are arguing they are protecting consumer’s right to privacy (which generally seems to have a lot of support) yet on the other hand they will still show you the search query if you are one of their legion of paying AdWords customers!
The Google Webmaster team argue that you shouldn’t worry about the change, as you can get access to all your search query, impression and ranking data in Google Webmaster tools and this is now integrated with Google Analytics. In GWT you can get:
· The top 1000 daily search queries and top 1000 daily landing pages for the past 30 days
· The impressions, clicks, clickthrough rate (CTR), and average position in search results for each query, and compare this to the previous 30 day period
The only trouble is that the reliability of this data has been questioned by many commentators:
Tom Critchlow’s blog on Distilled
In any event, I can only see data in GWT up to 17th October for sites that I know have been impacted which is no help yet!
The second problem is that this GWT data is not available in the Google Analytics API or the GWT API (which is rudimentary to say the least).
I’m not going to weigh into the debate just now….my primary concern is how does this affect our ability to get accurate data to our clients and our ability to make decisions on that data.
The big unanswered question so far is;
“What is a minority?”
What is the percentage of organic keyword visits that are affected and that will now show up as (not provided)?
Danny Sullivan on SEJ reports that Matt Cutts from Google’s search quality team has said that it should be in low single figures; but as any decent golfer (or web analyst) will tell you there’s a world of difference between a 1 and a 9!
So we’ve done our own research from just one of our webmaster and GA accounts, to try and help you get a picture of the impact of this change and to help you change your reports, analysis and insights accordingly.
We will update this analysis regularly until we start to see a strong pattern emerge; so feel free to subscribe to our email newsletter or RSS feed.
(not provided) average is (currently) 2.82 % of Organic Keyword Visits
Across a sample of 140 websites analysed (this research base will grow by the day) since 18th October to date, the average number of (not provided) queries expressed as a percentage of organic keyword visits was 2.82%.
But this does vary by site and we have seen this percentage as low as 0% and as high as 33.3% – see graphs below.
OK here’s the story so far. On Day 1, (18th October 2011) we saw 39 websites affected, with one site having as many as 3% of natural search traffic affected.
The average (not provided) percentage was 0.64%. If you look at unique sites on the 19th October this average rose to 3.52%.
So far if you look at unique sites on the 20th the figure has risen to 4.65%. It’s clearly too early to say if this upward trend will continue and where it will stop; but the average across all sites since this started is 2.82%.
For the 18th October to 20th October we are seeing an average across all effected sites of 2.46% overall. For just the 20th October we are seeing a lower percentage at 1.19%. The below chart displays the data over the full range we have been reporting on:
The biggest we’ve seen so far is 50% but the site had too few visits to be significant. But we have also seen a site with 325 of 1373 – 23.67% of visits affected!
It seems to me pretty likely that some marketers are going to be hit harder than others by this change. For SEO Software providers like ourselves I would wage that a larger proportion of our site visitors are likely to be Google + users or signed-in to Google (unless they are doing keyword rank checks or Google hacks!) than the norm.
The next big question is now you know the percentage of organic visits affected by this change what can you do about it?
I’m afraid the answer is very little! Your options are:
a) Get on the privacy bandwagon and complain that all query data should be hidden, including PPC query data, so it’s not just SEOs who feel penalised.
b) Get on with it; and adjust your SEO reports accordingly. Look at your keyword report in Google Analytics for a period and calculate your total number of organic keyword visits from (not provided) as a percentage of your total organic keyword visits. This will give you your percentage of signed-in Google users visiting your site through Google SSL search.
The only trouble is when you start to look at this organic data at a keyword level. For sites with very low numbers of organic visitors there’s little point applying a percentage uplift at a keyword level, especially for long tail terms.
The best way of analysing this is to track visits per keyword over time. So the good news for our users is that next week we will enhance our keyword visits data in Analytics SEO and show you a chart of organic keyword visits, as well as the current table of data showing organic keyword performance data since the start of your SEO campaign. We’ll also show you an estimated traffic based on your percentage of visitors that are signed-in.
As Pete Young points out in his blog post, I too fear that the impact of this change is only going to grow as other Google assets such as Google+ roll out.
In the meantime, please feel free to add your analysis, your insights and findings and your suggestions for further research.
If you have your own research then please let us know, we’ll happily publish it here – if it’s robust 😉 or link to it.
By: Laurie OToole