Web Statistics are not all that you thought?.

Benjamin Disraeli was heard to have said that there were lies, damn lies and statistics, and certainly when it comes to the internet this phrase can certainly be said to be true, especially as few, (or none) of the agents delivering these statistics agree with each other

Log Files, Google and Flightpath Analytics

I use three sources of data when checking on what is happening on my customers sites, and can say that in my experience none of them agree with the other, something which makes deciding what to believe and thus what to do very difficult indeed. Google even disagrees with itself, the number of clicks being charged for (as stated by the Adwords Engine) being different from the number that their own Google analytics tracks.

But apart from the numbers, does Google Analytics provide all the information needed? I've cross checked data with Flightpath Analytics and have found that Google, as far as pay per click visits are concerned, reports on the adwords adgroup that was "activated" and does not tell you what the keyword phrase was that actually typed in (and that thus caused the advert to be triggered), data that is vital when deciding on the all important negative keywords in an adwords account.

Time Spent on Site?

Then there is the data reporting on the time on site. I've always doubted the validity of this particular set of information, based as it must be on the time between clicks on a site. Now whilst this data works reasonably enough while the visitor is hopping from page to page on the site, what happens when the visitor leaves the site?, just how accurate can the data be for this last page view. People have told me that the systems use a time out of 20mins if they can't tell what has happened, others that the systems don't count the time on the last page. For myself I'm not convinced, and thus view this data with some suspicion...

Bounce Rates - Straight Exits from the Site

Bounce rates, a measure of how many times people visit pages and leave straight away are another well used statistic. This data may well be reasonable, but I do have some problems with some of the figures delivered by all systems, in that they all report pages well inside sites as being the entry point for a visit (that then results in a bounce). Now while this is quite feasible, in many of the cases that I have documented the data looks very suspect indeed.

A Few Reasons to Treat it all with a Pinch of Salt

There are a few reasons why the first visit to a site appears to be to a page deep inside a site. The most obvious is that the page had gained a good ranking on a Search Engine. Other reasons include a change in the IP address of a visitor, or by a timing issue where the visitor goes to one page and then another and then has a cup of tea etc and thus after 20 mins "times his visit out". In such cases if the visitor looks at just one more page and then leaves, they will be considered a new visitor that has "bounced", the page in question being by then deep inside a site. This and the fact that visitors can save an address in their favourites etc can mean that they can arrive in, and thus potentially bounce straight out again, of just about any page in a site. However, I am still wary of this data, which is a shame.

Final word on Google Analytics, Flightpath and Pay Per Click

The way I look at this data is to treat them all as a help, but I must stress that they cannot be totally relied on, and when it comes to Google Analytics, please be aware that the "paid" keyword report section, provides data that is not referring to the keywords being typed in, but the adwords adgroup name that has been triggered. Thus if you want to know which keywords should be considered for negatives, you'll need to use your log files, the "Search Query" Report in Adwords, or something like Flightpath to track what is really causing your clicks.




Graham Baylis
Internet Marketing and Promotion Specialists
www.I-O-M.com
"Making sure your needle is found in the Internet Haystack"