Google Analytics as a reason for frequent changes

As soon as LJ allowed Google Analytics code to be installed on the diary pages, I, of course, immediately turned on the counter. A very interesting feeling - as if the fog of war had cleared, it became clear where visitors came to my blog and what interests them (that's why we love web analytics, right?).

One of the most interesting reports on the site that I draw attention to is “Keywords”. These are requests from search engines that have found a website: some of them are just funny, others can help to understand the needs of users and think of what to write about. In short, the report is very useful.

At one point, I discovered that the most popular word that comes to my blog from search engines is “cheating on my wife.”
')


Scandals, intrigues, investigations! Of course, I went to look for myself in issuing Yandex on this request and did not find it. But the following explanation came to mind was true. I was able to understand what happened because I learned the properties of Google Analytics well, for example:

1. Google Analytics records traffic sources by visits, not unique visitors. If the same user comes to the site from Yandex three times, the search engine will count three transitions.
2. The transition of a user returning to the site without a referrer (visiting from a bookmark or from the address bar) is assigned the same source that he had before. For example, if I first opened the site using an advertising link, and then saved it in bookmarks and reopened it the next day, this visit will be counted as the second transition in advertising.
3. Transitions from subdomains of search engines do not differ among themselves: for example, a user who has come from a Yandex search in pictures is indistinguishable from one who has passed from a Yandex search in blogs.

So what happened with LiveJournal? Very simple: I remembered that a couple of weeks before that I was looking for an entry in my own diary containing these words. Of course, after that I went to read friendslite several times. And it turned out this:

1. Every my visit added to the statistics of visits.
2. Since I came from the bookmark, Google Analytics “remembered” that I had come from Yandex before, and considered it a new transition by keyword.
3. Since all types of searches from the Yandex domain are considered by Google Analytics as yandex.ru, I could not determine that it was the blog search that was used.

What can be advised to users of Analytics to avoid such situations?

1. If it is critical for your site, what kind of search sub-service visitors use (for example, Google’s regional domains or additional Yandex services), they must be added to the counter code forcibly. On this occasion, Daniel of Azov wrote an excellent article.
2. If possible, filter the transitions of company employees, the site owner and other “friends”. This can be done by filtering by IP addresses, but it will be even more accurate to add to the site code a condition according to which the counter is simply not shown to the administrator. The same thing could be done in LJ.
3. Always remember that statistics are counted by visits, and not by unique users. Then you will not be surprised, for example, that Analytics will count more transitions in contextual advertising Yandex.Direct than Yandex itself.

I really hope that LJ will expand the installation of the counter: it will add variables to determine the author of the diary, as well as the code for highlighting the Yandex search on blogs in a separate search engine. And recently I was visited by the request “the last session of 2010 of the Lipetsk regional council”, and I don’t even know what to think!