Timing Average Time on Site

[Frank_Rizzo]

I'm intrigued how loganalysers calculate this.

The log file records when an item is accessed (lets say grabbed) but it doesn't log when that item is released.

So how do they do it?

PS
Rather than pay hundreds for a full blown log analyser product isn't easier to put the log files in say Access and run queeries on that?

[broniusm]

I only know of difference between timeStart and nextTimeStart.
index.php 05:00 =>
contact.php 05:12

gives 00:12 on the default page..
Where detailed clientside info is important (like this one mock e-retail site I did for a university research project), I used document.onload to init a page timer and .onbeforeunload in IE to calculate the diff and store that in a cookie. Each subsequent page ate the cookie and stored it in a session variable. Finally, I had the luxury of a postsurvey which popped up afterward, giving the opportunity to read the session and store it in the db in one fell swoop.

cheers
-bronius

[Frank_Rizzo]

Yes, thats fair enough putting some code on your pages, but log file analysers I've tried all show some kind of vistor stay length figures.

Some show who stayed the longest, some show the average for all visitors, average page view etc.

I can understand that it is possible to calc the visit length of a page if a visitor views different pages in succession

view page A, start = now1
view page B, length of view page A = now - now1

But how can they work out the average view of say a view of the homepage and nothing else? How do they know when Elvis has left the building?

[wkitty42]

i know that this is a bit late but i'll offer my comments anyway...

i've got my site set to issue a "tracking cookie" for each user... this cookie stays active during the course of their visit... i then have a special cookie log that records the cookie and the URLs requested... traditionally, many would call this a "site navigation log" since it shows a visitor's path thru your site... in any case, its a simple matter to take the last entry and subtract the first entry from it to determine the length of time they spent on the site...

you can also set an abirtrary time length and then follow a specific ip or domain entry thru yout normal access log... in my case, i have webalizer set to record visits of 30 minutes or less as one visit... if they go 35 minutes, then i see it as two visits... if they go 10 minutes, 10 minutes and 10 minutes then i see it as one visit... yes, this method is not as accurate as the cookie log one... sometimes you have to work with both to get more accurate...

i won't mention that the cookie log method breaks in a gazillion pieces when visitors do not allow your cookies and do not return them when they are requested... thems the breaks, though... the web wasn't designed for tracking of stuff like this...

[analyst]

Frank,

Since people leave pages opne without looking at them, shift windows, move to other sites, and turn off their PCs, these "time" stats are meaningless, at least in an ABSOLUTE sense.

In a RELATIVE sense, they might have some meaning. I spotted one page that had notably shorter time than its peers. Sure enough, it was shorter and skimpier than the others. I should add content to it, I guess. On the other hand, the pages with the longest times might deserve to get split into two pages.

P.S. For free, you can get Analog, the best log analyzer going. BTW, the Analog developer doesn't believe in "time" stats, so they aren't there.

[statomatic]

Frank, as an author of such software I can tell you how it is calculated using server logs. You can only calculate it as a difference between first and last requests. Time spent on last page cannot be recorded as you don't know when the visitor leaves it. Obviously this is not 100% accurate, almost none of the reports are, but on a site with a lot of traffic can be used as a reference point. For example last month average visit duration was 2 mins and this month 3 mins, you'll know that people stick around longer. Hope this helps. Cheers.

[mild green fairy]

I've always considered the whole 'average stay length'/stickiness thing to be a totally bogus metric. What's the distinction between someone wallowing around the site unable to find what they're looking for and someone who's actually browsing through some useful pages? Similarly someone who gets in, finds what they want inside 10 seconds and gets out looks just like someone who's take a quick glance and decided you're not for them.