The Horror. Precious data taken from our fingertips.
On the 16th of October, Google launched secure search for anyone logged in to their Google accounts, which mean analysts everywhere started seeing (not provided) as a keyword in their reports, even though we clearly hadn’t SEO’d for it (yet).
A full discussion on it, and some ways to analyze its impact can be found on Avinash Kaushik’s blog. So read that first, and report back here when you’re done. The post below is the results of a thought process I came up with after reading his post, so it’s important to get his background. He’s also a lot smarter than me, so that’s a second good reason to read his first.
But I digress.
Google Webmaster Tools and SSL Search
As soon as the option became available in Google Analytics to connect to my Google Webmaster tools account, I took advantage. And I think it was fortuitous in this particular instance. What I propose here is that we use the new SEO Reports as a benchmark to check for changes in our incoming keyword volume that might be hidden in (not provided).
Let’s have a look at some results from a durable goods manufacturing website.
Overall Search Visits
We find that organic search traffic is pretty steady. So no reason to assume we have any problems that will affect the test.
Google Organic Search Visits
Ditto for Google specific organic search volumes. I think we have a winning sample set (albeit small compared to many websites out there).
The Dreaded (Not Provided) Organic Search
Oh boy. Here we go. We find that roughly 13% of our Google Organic search is affected.
Some things to note:
- We are not a tech company, nor do I suppose we have a particularly techy customer base, considering the product set.
- Very few of our leads that come in are from Gmail addresses. We get mostly named domain contact emails.
So. Where do we go from here? What keyword values make up that portion of search that now says (Not Provided)? How do I measure my SEO efforts and content marketing?
To me, one of the places I want to look is in our SEO reports in Google Analytics. We shouldn’t see any dramatic changes, given our steady traffic, but you never know.
The Queries Report
Ah. Nope. No changes there…… These are approximations of course, but it doesn’t get any more steady than that.
So next, let’s take out image search, and see what keywords changed if any.
Web Property Only Search
So here’s the info for our best converting keyword. We see impressions remain even at 1,600, and our results position remained about 5th on average, but 30% of our clicks have disappeared. This should immediately raise a flag. Again, these are rounded estimations, not “real numbers”.
So what we need to do is check against a few other things, and see if we can prove that some of those 30 missing clicks ended up in (Not Provided).
Let’s look at the keyword in the Search portion of Google Analytics and compare.
When we look at the “real numbers”, we find the same percentage of searches missing. So it looks like we may have found about 10% of our (not provided) searches. There’s a few more checkpoints we can use though.
To accomplish this next step, I used an Advanced Segment : Google Organic Traffic Only. I then went to Landing Pages under Content.
Landing Page Analysis
Here we see (Not Provided) as the first most popular phrase for our Landing Page (grrrrrrrrrr…..). Our best converting keyword is number 2 on the list. We CAN infer that 30 of those 158 came directly from our best converting keywords. That’s roughly 20% of the (not provided) traffic.
We also know that it’s our best converting organic keyword, so let’s have a look at that.
Testing SSL Search by Conversion Rate
Lastly, I wanted to look at Conversion Rates. I know the keyword in question converts at a 25 to 30% better clip than the rest of our organic traffic. (Want to guess how I spend my SEO time?). For the graph below, I only look at the true conversions. Those that requested quotes, or requested call backs/contacts. Here’s what I found:
Oh brother. Not only is the SSL search catching 30% of my best organic keyword searches, but apparently it’s devouring a lot of rare long-tail too, that converts at an even BETTER clip. So I will need to do a LOT of digging over time to find out what those are.
Some Takeaways From My SSL Search (Not Provided) Experiment
I think we can still get back some of our data through analysis like this. Some things I’m going to do in order to find that long tail in the future:
- Examine Actual Keyword data from our AdWords account and see where those 9% converting keywords might lie.
- Create another Advanced Segment that studies Bing Traffic to that Landing Page, which should be similar.
- Write angry tweets at Google Developers until they revert.
- Build a MUCH deeper historical keyword database to compare against.
How about you? How do you plan on getting through this new event horizon of web analytics? Please share below!