Yet Another Way to Reclaim Your (not provided) Data

Since Google started rolling out its SSL search for logged-in users, SEOs have been scrambling to find ways to reclaim some of the lost data. Our industry relies heavily on keyword data gathered by web analytics software. For someone such as myself, who thrives on web analytics data, seeing the accuracy and usefulness of analytics data eroded further causes an almost physical pain.

There have been many very good approaches with custom reports and advanced segments to gain some measure of insight in to the (not provided) traffic. There’s a very good post by Rachael Gerson on SEER Interactive which talks about looking at the landing pages where (not provided) keywords arrived on, and David Harry has a thorough examination of (not provided) on Search News Central.

Today I will add a new approach to (not provided), which I hope will further help undo the damage SSL search has done – and continues to do – to our keyword reports.

All the following is based on Google Analytics, and I sincerely recommend you create a new profile for these filters and segments so that your standard reporting is unaffected.

To start with, we need to implement a filter in Google Analytics that’s been around for a while. This filter extracts the actual rank of a keyword when a user clicks on it on a Google SERP, and attaches it to the site’s organic keyword report in Google Analytics.

First explained by André Scholten on Yoast.com, it’s also been thoroughly detailed by Bryan Casson on YouMoz. Below I’ve included a screenshot from Bryan’s YouMoz post, but I won’t repeat all the details here as I’d merely be duplicating the work done by André and Bryan.

So please read the posts linked above to see how it is done, and be sure to implement both the keyword extraction filter and the ranking display filter in the right order on your Google Analytics profile (i.e. first the rank extraction filter, then the display filter).

Next you wait a little while for Google Analytics to start populating this new profile with the filtered data, and soon you’ll be able to see ranking data of keywords in your organic search reports:

keyword-positions

Now it’s just a matter of looking at all keywords for a given rank – including the (not provided) keywords, combined with the keyword’s landing page. Filter the report for a given rank, and sort it by landing page, and you get some mighty useful information:

not-provided-landingpage1
not-provided-landingpage1
not-provided-landingpage1
not-provided-landingpage1

If the (not provided) keyword has the same rank and leads to the same landing page, chances are it’s the exact same keyword.

Of course for some popular landing pages, the keywords and (not provided) ranking data are not quite as easily correlated:

not-provided-landingpage-list

Still, by using these Google Analytics filters and comparing (not provided) traffic to the keyword ranking data and landing pages information that is available, you can make pretty accurate educated guesses about which keywords are contained within your (not provided) black hole. You can rejig and combine this data in any way you want to extract maximum value from it, and hopefully recover some of the lost accuracy in your SEO reporting.

If you like this sort of custom analytics reporting, be sure to check out the new Custom Report Sharing for Google Analytics website, which contains an ever-growing list of awesome custom reports, filters, and segments for your web analytics pleasure.

About Barry Adams

Barry Adams is one of the chief editors of State of Digital and is an award-winning SEO consultant delivering specialised technical SEO services to clients worldwide.

25 thoughts on “Yet Another Way to Reclaim Your (not provided) Data

      1. I think it’s a big guess, but this looks even more guessed than regular Google Adwords Suggestion we all love so much.

        1. I don’t know Roy. When you have a (not provided) keyword with the same rank and same landing page, and there might be a similar bounce rates and other relevant metrics… I think that makes it a helluva lot more accurate than the laughable AdWords keyword tool, imho.

  1. I’m not sure either. It will always be a guess. You can try to make the best guess possible, but you’re never sure, which can make it misleading for your analysis.

    The question is, do you really miss that much data that you need to take these kind of actions? If bouncerates are high and you don’t know which keywords the visitors have used, you might try to focus more on internal search and let the search form on your website stand out more. The data you gather with that is way more valuable. 

    1. Agree to some extent – it will always be a ‘best guess’. But it’s better than having no idea at all. And at least this way if there’s some unexpected behaviour on specific pages with a lot of (not provided) entries, you have a chance of analysing it with a bit more accuracy than if you’d just rely on GA’s standard reporting.

      1. True – you will have more data to analyse when you discover certain problem areas. The upside of this solution is that when you see high bounce rates on a page, you might be able to identify which keywords are causing this if you have more visible keywords with the same position and bounce rates. 

  2. Thanks for the mention! I can tell you that, on my side, I’m moving away from the granular keyword reporting and trying to focus more on landing pages. I’ve been using an advanced filter to pair the (not provided) keyword with the landing page, rather than with rankings, for this reason. We’re all evolving in our reporting, it’s great to see what you’re doing there!

  3. Pingback: SEO | Pearltrees

Comments are closed.