Last time out we looked at the hidden value of (not provided) in that we now know what percentage of visitors to your site are (potentially) seeing personalized results and at very least, logged in. This time out we will look at little closer at some data to get a sense of understanding it all.
What is this jibberish about (not provided) you ask? It’s related to Google’s changes into encrypted search which now only returns (not provided) in the keyword reports in Google Analytics. For more, use the link above and read the first post in this series.
What’s the point of it all?
First off we need to establish what it is we want to try and wrangle from the data. Some starter questions might include;
- What percentage of the search traffic is logged in to Google?
- What’s the user demographic? (compared to known sources)
- How are they engaging on the site?
- How are they converting?
- What social actions are they taking?
- Which pages have the most NP visitors?
- What is the value of NP visitors?
Avinash put it well with,
“We want to know more. Who are these people? Are they people I should care about? Not care about? And what kind of search queries are these? Brand? Non-brand? What else? “
Our goal is to establish the user type and value of people coming to the site via a logged in state. We might be making changes to the site, social programs and more if it seems warranted. But first we need to establish if that is indeed the case.
Getting started; Build some segments
The first thing we’re gonna want to do is build a few handy Advanced Segments to give us easy access to the brand, head, tail and (not provided) traffic. Just click on the Advanced Segments tab and then create a new segment.
The firs one you can just name ‘Not Provided’ and it’s the easiest one. Simply include exact match keyword, like so;
Yer done. Save it and then create another new one.
This time we’ll call it ‘Head Terms’ and get a little regex action going on. You can name this one ‘Head Terms’ and use the following regex; ^s*[^s]+(s+[^s]+){2,}s*$
What we’re doing here is showing keywords with 3 or more words. Remember to replace ‘brand terms’ with actual brand terms for your site. Save the segment and create a new one.
This time we just want to do the inverse of the one above, to include keywords with 3 or more. Name this one ‘Long Tail Terms’ and follow the same method as above, just change the first element, (one with the regex) from ‘exclude’ to ‘include’. Leave the rest the same. Then save it.
The only thing we have left now is the brand terms. All you need to do here is create a new segment, ‘include’ your brand terms (I use regex) and we’re ready to go. We should now have advanced segments to isolate; long tail, head, brand and (not provided) data.
Custom report; Not provided
The next thing we’re going to want to do is create a custom report for the (not provided) data. Go to ‘Custom Reports’ and click ‘New Custom Report’ and add the following;
This will create a report that segments all the NP data and has a drill down to actual landing pages that are associated with it. This should look something like this, (after drill down);
Custom report; Search performance
The next report we want to create is a search performance report. This one is actually handy for a few reasons, not just for our current need, (for NP analysis). I actually use this one on my dashboard as it’s a handy set of data to have easy access to.
As usual, go and create a new custom report and call it ‘Search Performance’. Set it up like so;
For the ‘Include’ add all the search engines that you want to track. And in the ‘Exclude’ we want to remove elements such as ‘plus’ ‘mail’ and other traffic coming from Google/Yahoo/Bing that aren’t actual search traffic referrers.
Add (not provided) widget to dashboard.
We also want a quick way to get a sense of how much traffic is actually coming via encrypted search. For this, we want to navigate to the ‘Not Provided’ custom report we made. Then click on ‘Add to Dashboard’. Then use the following;
Once you have it on your dashboard, we need to edit it…
That should now give you an easy way to keep track of the level of NP at a glance from your dashboard. We have now established the first question; how much of the Google traffic is coming from encrypted search?
Analyzing (not provided) traffic.
Now that we’re ready to roll, it’s time to do some traffic analysis to get a sense of who the visitors are that are coming to the site in a logged in state, via Google. Across the many GA accounts I manage the rate of NP varies wildly. In some case under 10% and in others nearing almost 30%, (madness). That’s our first clue.
Given that (not provided) is related to logged in users and logged in users have the potential for personalization, (behavioural, social, geo) it is important to really get a sense of what is going on.
Query type breakdown
Query classification is what most SEOs know as the query types; informational, transactional, navigational. One interesting observation to make is how they stack up to the NP. For that navigate back to your search performance report and then add the segments; Head Terms, Long Tail, Brand and Not Provided.
It should look like this;
What do we know?
- Visits ““ it is most closely related to the head terms and is good for about 15% of total traffic.
- New visits ““ the NP is roughly in the same range as the other query types
- Time on site ““ the traffic most closely resembles long tail traffic and the lowest of the group.
- Bounce rate ““ once more, least effective of the group and resembles long tail traffic
- Pages/Visit ““ the (not provided) is the least engaged.
- Conversion rate ““ again, our weakest segment.
This kind of analysis can start to give us an idea of the how these visitors are engaging the site as well as some potential insight into what kind of terms these visitors might actually be using. In this case, it seems that a good chunk of them are likely long tail terms.
Regardless, we now have a sense of the value of these visitors. So, what else can we learn?
Pages getting (not provided) traffic
Staying on that report, we can now click on ‘Google’ to get a page by page breakdown for the same query data. It will look something like this;
This now enables us to get a deeper sense of which pages or types of content are getting the (not provided) referrers and the related performance. In the example above, we note that the home page traffic from NP is actually quite relevant. This helps clarify the original hypothesis that this was weak traffic. It can vary by page.
Of course each site is going to be different. The main idea is to identify which types of pages are being effective from the NP traffic. We can then get a better sense of exactly who they are and which areas of the site perform best.
The Social Connection
As we’ve discussed along this journey, the (not provided) data is about logged in Google users. And of course a logged in user means a user that is potentially seeing personalized results, including social graph elements.
As such, we also might want to add a ‘Social’ data point into the mix with your custom report. They look like this (in edit area);
As always we can then better understand the social sharing activity for the NP visitors, compared to the other classification types. How social are these logged in users compared to non-logged in? This might go towards the types of content your producing, the social optimization of the site and changes to social program levels.
Putting it to use
And so my friends, what was the point of the exercise? At the very least we now have some tracking in pace to better understand the NP traffic. It can lessen some of the anxiety for yourself and clients/managers who might be freaking out. At least.
At most, we now have a better sense of the type of visitors, what they’re doing on the site and how they’re converting. While there are many ways we can use this, one is certainly in the area of content targeting. Get to know which pages and types of content are doing best with these users. There are certainly some patterns which might be repeatable. Treat them as you would any user type and work towards better serving it.
Does it give us back our keyword data? Well, no. But we can at least seek some deeper understanding of these potentially personalized referrers. To better understand personalization elements, see my recent Search Engine Watch post; Search Personalization & the User Experience
I have only recently devoted time to looking closer at the (not provided) data and as I have more insights/ideas/additions, I shall post them here. Be sure to come back often!
Until next time, play safe!
