‘Keyword Unavailable’ arrives in Adobe SiteCatalyst

Those of you who spend your days staring aimlessly at Search term reports in Omniture may have noticed a new ‘feature’ arriving in the last couple of days. Yes, that’s right, ‘Keyword Unavailable’ has hit SiteCatalyst reports as of the 11th November. Ignoring the obvious conclusion that this was the beginning of the end of the world, this was actually a long awaited addition to the reports that will be invaluable.

Keyword Unavailable as a value in your search reports

Of course I’ve gone over the reason that is is appearing in your reports several times in the last couple of weeks, but for those of you who are uninitiated, here is a potted history:

Firstly Google updated the way that their search engine worked by making it so that if you were logged in then you were going to be using a secure search. Using a secure search means that Google won’t pass a referrer value if moving to a non-secure domain. This happens most of the time, so most people wouldn’t have seen a referrer. Google, of course, has fudged it so that you still know that the user is coming from Google, but you just don’t know what they searched for any more. This resulted in a (not provided) being a value that you’d see in Google Analytics when a user came from a secure search.

All the other web analytics tools were caught by surprise by this move by Google and didn’t have a system in place to do anything about it, so Omniture suggested using a bit of a fudge and others suggested changing the s_code. This was widely derided (if you count me as ‘widely’) and Omniture came up with their own solution which they implemented last week – you will now see ‘Keyword Unavailable’ in your reports (I wish they’d used (not provided) like Google did).

In fact I later pointed out that this whole change has introduced a bias into your search data that didn’t exist before. This is especially true if your data is getting more of these searches, but has it? Well you can see that in my example one of our clients is getting about 1% of search visits whereas another is getting about 4.5%.

But this is just set to rise as more and more people use services like gmail and Google Plus (or any other online Google service). In fact over at eConsultancy, one of their most popular pages is looking at about 33% of its US traffic coming from (not provided). That is a real serious sample bias. Especially seeing as second on that list of keywords only produces about 3% of their keywords. They are getting serious long tail traffic on that page and they won’t really know about it from now on.

Equally on whencanistop (another blog that I write) I noted that in the 4 years that I had been writing:

One of the advantages of writing lots of long blog posts is that you get lots of long tail search traffic. This is particularly true of this particular blog. 8,757 different search terms have generated 14,480 visits. It’s difficult enough to work out what is going on with a few keywords, but when half of your visits from search engines come from a phrase that was used to visit your blog just once, it becomes increasingly difficult.

This month so far 16% of my search terms have come from (not provided). This is a serious hamper to my long tail keyword analysis which was hard enough to start with. I’m going to have to use my webmaster tools data doubly so (although that has limited scope at lower volumes because of the sampling it does) compared to how I’d normally do it.

It is hard times indeed for the analyst.