Living in a world of 100% ‘not provided’


Living in a world of 100% ‘not provided’ ...

As many of you are aware, we are fast approaching a world of 100% ‘not provided’ organic traffic source data. Over recent weeks we’ve noticed not provided traffic increase at an alarming rate and we are now seeing an average figure of over 75% ‘not provided’ data.

In conjunction with this recent industry blog posts have fuelled speculation that over the next few months Google is rolling out SSL protocol to all searches – regardless of whether a user is signed in or not. Due to secure search encrypting search queries marketers are now seeing (not provided) in place of the actual search query that referred the searcher to the site.

You may remember that way back in 2011 Google first introduced SSL encryption for searchers signed into their Google accounts. More recently they introduced the same level of encryption for searches searching in the Chrome omnibox.

Other browsers such as Firefox and Safari were also quick to upgrade to SSL search, further increasing the number of ‘not provided’ organic search site traffic.

The new organic search data landscape

The reality is that it looks like the days of seeing our beloved organic search keyword referral data in GA are over. At the same time we shouldn’t be crying over spilt milk – we’ve all known for some time that this day was coming.

Whilst it does make analysing and understanding things like page performance and the impact of optimisations more difficult, there are still a number of analysis options available to us.

Options for getting round the change and viewing data in a different way

Firstly in Webmaster tools you can export up to 2,000 referring search queries a day for a 90 day lookback window. The more technical savvy can automate this download process using Python.

The webmaster tools data does have its drawbacks. Large retailers with a large longtail are going to find the 2,000 search queries limit a day woefully inadequate. (However, personally I’ve found it frustrating that you can filter search query information by search vertical and location in WMT but in GA this reporting functionality is not available out of the box.)

From an analysis perspective I’ve always felt that the top landing pages report in Google Analytics is underused for evaluating optimisation success. All you have to do here is add a non-paid search advanced segment. Those of you familiar with regular expressions can also utilise these in the advanced search box to isolate particular groups of pages and further enhance reporting.

There are also ways to create a custom view in GA and use advance filters to aggregate page data to understand performance and optimsitions effects. Going forward we will need to think differently and use more of this type of analysis to see the trends.

Internal site search data is also another untapped resource, and if used correctly, is a goldmine of useful facts and figures that can be used to power strategy.

Referring search query data also had its flaws. In the world of personalised results with the exception of search referrals that contained one of the six question words it was partially difficult to understand the intent of the user. It’s also important to consider that referring search query is also a session based metric.  In this example think of a site visitor who visits your site multiple times, on different devices through different channels. Perhaps focusing on an individual referring keyword here isn’t the most sensible use of your time?

New advanced segment UI could make us better analysts

Perhaps most refreshingly the new advanced segment UI and features will allow us to answer some of questions in a different way. New features such a cohort analysis and sequence segments are now allowing us to measure the user rather than the individual visit – a far more powerful option.

This shift is important and part of an overall trend. There is no doubt that keyword referral data is extremely useful when available but now that we are approaching a world of 100% of ‘not provided’ data we just need to think differently about how we analyse the data we do have. Finally we might end up concluding the change is not that bad for marketers after all!


Contact Us

Do you have a challenge for us to solve?

get in touch