(not provided) Keywords
Croud’s SEO Manager Joao da Costa, takes a look at the ways Google keyword (not provided) is affecting our data analysis capabilities and suggests approaches we can take to overcome it.
Google’s introduction of SSL Search back on 18th October 2011 has certainly impacted the way we measure SEO campaigns and justify spend. But what is it and how this is affecting us?
Google is committed to improving security, protecting privacy and building tools to give more control to users. Based on such principles Google developed searches in https (secure), which gives users a certain level of security and privacy. If a user performs a search while logged into their Google Account (this includes Google+, Gmail, YouTube and so on) it will have the following URL:
What does Google say about it?
“As search becomes an increasingly customized experience, we recognize the growing importance of protecting the personalized search results we deliver. As a result, we’re enhancing our default search experience for signed-in users.” Extract from the Google Blog.
This basically means that Google will hide the keyword information from webmasters and marketers. The Source and Medium will display as google/organic as usual, however keywords will display as (not provided).
But what does it really mean for SEOs?
Without the granular level of keyword analysis as we used to have, analysing our campaigns becomes more complicated and work some. Common barriers we find are:
- It reduces the understanding of SEO campaigns: when optimising a website for a range of non-brand keywords it is important to understand what keywords are performing well and should receive an extra level of attention
- It makes it more difficult to prove ROI of SEO: being able to attach a sale/lead to a keyword helps to prove investment in SEO, demonstrating how SEO is generating sales
- It minimises the potential to react to failing campaigns: not having full keyword data will reduce the chances to identify keywords which should be dropped to give space for a new keyword with higher conversion potential
There are, however, other managerial level problems being created because of this, such as:
- SEO is usually not fully comprehended by management: sometimes it’s even seen as a black art. Reducing the level in which SEOs can prove campaign ROI will make SEO much more speculative and less black and white
- Campaign spend allocation: companies that are sales driven, tend to allocate high budgets to activities that bring ROI, if non-brand campaigns are not able to prove its effectiveness and its value, it can bring the budget to a halt
- Team development/agency: similar to the example above, if a campaign is not receiving more budget, it will also not receive extra reinforcements by increasing the team size or even getting green light to have an agency upping the SEO game
How much data (not provided) is hiding from SEOs?
We have recently carried out a study to identify how much data is being hidden from websites’ Organic Traffic.
For this study we analysed 10 websites from different industries and also got a mix of B2B and B2C.
Our key findings are shown below:
Since October 2011 the number of (not provided) keywords increased gradually reaching now a total of 40% of the Organic Traffic.
In March 2012 the number of (not provided) keywords increased due to Firefox built-in google search started to use https setting by default and therefore hiding keyword information.
We were also able to identify that the % of (not provided) tends to be higher for B2B brands and for B2C brands that sell technology related products. Other B2C brands that sell products like gardening have shown the lowest % of (not provided) keywords.
Such increase in (not provided) can be directly linked to the increasing popularity of Google+ users, Google Chrome and other Google products. We can also identify that the more people who actively log in in their Google Accounts, the higher the number of (not provided) will be.
According to notprovidedcount.com, (not provided) will hit 100% by 26th May 2016, if current growth rate remains the same. This is a good indication of how quickly (not provided) has been growing recently.
Yes, ways to track effectiveness of our SEO campaigns are becoming more challenging, but how can we analyse data effectively now?
How can we overcome such barriers?
Below we are listing some ways in which you can overcome some of the (not provided) data restriction and add data back in by applying predictive assumptions:
Brand and non-brand % of traffic calculation
This is a straight forward way to predict what the total brand and non-brand traffic actually were. Create advanced segments for Brand, Non-Brand and (not provided) keywords. This will help you obtain a percentage of traffic that each type of keyword is receiving, e.g.
This indicates that Brand keywords received 38% of visits while Non-Brand received 72%. We have to now split the (not provided) keywords according to brand and non-brand keywords traffic percentage. We will then have:
Brand Keywords: 337
Non-Brand Keywords: 551
Total of Organic Traffic: 888
Pitfalls of method: such method is a straight forward way to assume what (not provided) keywords are, however it assumes that (not provided) visitors have the same split as the current brand and non-brand keywords.
Separate (not provided) based on level of engagement
A more thorough approach than the previous one, however it requires a higher level of data analysis and analytics skills.
Analyse the level of engagement brand and non-brand keywords have and then allocate (not provided) keywords with similar metrics to their respective group.
This methodology is a bit more complex to implement, however it provides a more accurate allocation method rather than only considering % of traffic.
Pitfalls of method: such an approach can be complicated to generate, and engagement may oscillate on a monthly basis, therefore it requires an on-going analysis and less automation.
Top Level analysis of Organic Traffic
Another way that many companies seem to adopt is the top level Organic Traffic analysis. Such an approach does not separate brand to non-brand keywords, it just analyses both activities as one. Any growth seen in organic traffic is directly attributed to the SEO activities.
Pitfalls of method: this is a very poor method to base your efforts on as it doesn’t consider other relevant activities and external factors that might be affecting the organic traffic, such as: off line activities, PPC, brand popularity decline or increase and so on.
Landing Pages analysis
This has proven to be a common method used by agencies. Brand keywords usually concentrate in the home page of websites, leaving other pages to receive more non-brand traffic than brand (usually!). This allows SEOs to estimate that the majority of traffic received on landing pages are non-brand, therefore traffic increase in those pages will be attributed to the SEO campaign.
As it can be observed the landing page above receives 216 non-brand visits and only 3 odd brand visits. Such analysis can be automated by applying Custom Reports and/or Advanced Segments in Google Analytics.
Pitfalls of method: Landing pages have to be selected and analysed in order to ensure traffic is predominantly non-brand. If too much brand traffic is present, the landing page will not be representative.
Other more generic ways to analyse
Don’t forget that there are other ways to obtain information on your keywords. You can use Webmaster tools to help inform about search queries – Google says: “They can also receive an aggregated list of the top 1,000 search queries that drove traffic to their site for each of the past 30 days through Google Webmaster Tools.”
PPC is still a great way to obtain intelligence of what keywords are the most profitable. Such activity is not new as many SEOs use PPC data (which is quickly obtained and straight forward) in order to advise on what terms should be used in SEO which is a medium to long term activity.
Any SEO that has been in the game for some years can tell that SEO has been a constant changing discipline, in one way or another. Yes, it can certainly complicate our lives more, but when has Google ever said “let’s make SEO lives easier by implementing…”. Google’s first of the “Ten Things” they believe in is: “Focus on the user and all else will follow” –Google Philosophy.
Statistics are a fundamental area of data analysis and the more it is used, combined with analytics, the greater the potential to predict and anticipate changes. As Arthur Conan Doyle puts in The Complete Sherlock Holmes Book: “While the individual man is an insoluble puzzle, in the aggregate he becomes a mathematical certainty. You can, for example, never foretell what any one man will be up to, but you can say with precision what an average number will be up to. Individuals vary, but percentages remain constant. So says the statistician.”
As Google’s Evangelist Avinash Kaushik says about that analysis: “Go as deep as you can. Then, a little bit more.”