Google, Click Tracks and Bitstrike Log Analyser

Worrying differences in traffic reports

[designhaus]

The below are stats taken from 3 different programs. I have found some quite worrying differences between the three programs. Has anybody else seen these kind of differences ... has anybody else seen these kind of changes themselves?

Any comments and help to maybe solving an error I am making are very welcome.

Stats taken from Google, ClickTracks and Bitstrike Log Analyser

Hotel Website for the 7th to the 13th May

PPC Traffic for the 7th to the 13th May
Google Adwords downloaded report -- 76 clicks
Click tracks report -- 65 Clicks
Bitstrike Log Analyser Report -- 68 Clicks

Natural Traffic for the 7th to the 13th May
Click Tracks -- 243 visits
Bitstrike Log Analyser -- 250 visits

Car Hire Website for the 7th to the 13th May

PPC Traffic for the 7th to the 13th May
Google Adwords Report -- 11 clicks
Click tracks report -- 7 Clicks
Bitstrike Log Analyser Report -- 19 Clicks

Natural Traffic for the 7th to the 13th May
Click Tracks -- 1147 visits
Bitstrike Log Analyser -- 1128 visits

[edit_g]

Yup - I have seen similar stuff. Because each of the progs/applications tracks sessions differently - each one will show slightly different results. This can be to do with your settings (in the log analysis progs), how long cookies are set to last with each tracking mechanism etc... I really wouldn't worry - no two stats packages are ever going to match up completely. Hope that helps.

[Receptional]

I'd say that those figures are about as close as you will get from log based systems. The fact that they are similar to within about 5% on the natural stuff is fine. I can see you worrying about the Google paid traffic being disproportionally high, but unfortunately you do not have enough paid clicks to be able to see if this is Google Adwords trying to win one over on you or is just people trying to click on an adword and your server taking too long to respond, so the user goes elsewhere.

Naturally, Google won't be trying to upset you over 4 clicks, but also won't have time to investigate a claim either.

I would say that figures over 1000 will generally start show you trend or measurement errors if they exist. under that and the web is just too unpredictable.

Dixon.