Google’s encrypted search data is increasing in volume every month. Beginning in the US last October and expanding to the UK in March this year, logged in Google users now have their search data encrypted, leaving web analytics software blind to their keyword data. The words “not provided” appear in brackets in keyword reports for these encrypted searches. Then, in July Firefox 14 was released with encrypted search as the default for all its users, further increasing the amount of “not provided” keywords. Firefox has a 22% market share and I would assume a future version of Chrome will do the same, adding their 28% market share to the mix. In short, “not provided” keywords are on the rise and it doesn’t look like it’s going to change any time soon. Most SEOs have been aware of this increase throughout 2012, with many (rightly in my opinion) predicting it will rise to 80-90% of all Google search traffic. As such, SEO blogs are full of posts about how to get round the issue, minimise its effects or adapt to the new landscape. What I haven’t seen is a blog about the impact of “not provided” keywords on data analysis and reporting this year. The effect of “not provided” has been to combine many keywords into 1 in analytics and rather than do it suddenly, it’s gradually crept up all year. This means comparing search data to previous years or even within this year has become much harder. There are now a number of hidden pitfalls thanks to “not provided” data and they can all cause misinterpretations of a site’s data. All 3 of the below have fooled me and the self-administered forehead slaps have got louder with each new one. After the 3^rd time it happened to me, I realised it was time to get philanthropic and blog about it so you could avoid these perils.

1. The Long Tail Illusion

A metric we use to show an effect of SEO work is the growth of the number of keywords the site gets traffic from. As a site’s authority and content increases, so does the number of keywords, particularly long tail, it ranks for. Also, in diagnosing changes in traffic, looking at the number of referring keywords can help diagnose the cause. “Not provided” data is causing the number of referring keywords to drop because many keywords are being reduced to just 1: “not provided”. This can result in a misinterpretation of the data and the impression that the site is losing long tail rankings when it isn’t.

2. The Reduced CTR Fallacy

While it’s true that organic SERPs have been losing traffic share this year, this effect is exaggerated by “not provided” data in Google. All of a site’s keywords will have appeared to have lost traffic in 2012 because many of these searches will be encrypted and shown as “not provided”. You may be seeing a loss of traffic for a keyword that has held its ranking all year and reasoning this is because of lost CTR. It’s possible it’s just because many visits are now called “not provided” now.

3. The Brand to Non-Brand Ratio Error

A less common metric for search reporting is the ratio of brand to non-brand traffic. If you’re working on increasing the rankings of non-brand keywords, you will be expecting to see the ratio grow in non-brand’s favour. Again, “not provided” is coming along here to spoil the party. Arguably, you need to remove the “not provided” keywords out of the mix. But if you’re starting ratio is 80:20 to brand then it is brand that will lose the most share to “not provided”. It will appear that non-brand is increasing and brand is decreasing but this may not be the case. As I say, all the above issues have caught me out this year. They are regular metrics and you may even have them saved as custom reports in Google Analytics. I’d be very interested in hearing if anyone has actually made these mistakes, like I did.

A topic that had predominately been the focus of many debates last year was click tracking specifically in natural search arena. For those who are not familiar with the jargon I am referring to web analytics tracking. Typically all free or generic tracking tools work on the basis of tracking the last click. The last click is deemed as the last search and click through to your website by a user from a natural search result converting into a sale or lead. Normally everyone would think that it is okay except it is not! Here is why:

1. Google analytics only tracks the last click performed by a user as they arrive on a website and not the first (historically original) click. Therefore you can’t tell how a visitor originally found your site, making it difficult to prove conversions of non-brand traffic when users return via a brand keyword to finalise a purchase or make an enquiry.
2. There is no history on what the user had previously searched for or seen on the website.
3. There is no easy way of integrating a multi click solution

   Read the rest of this entry »