Beefy Boxes and Bandwidth Generously Provided by pair Networks
laziness, impatience, and hubris
 
PerlMonks  

Re: Why PM needs web stats

by aaron_baugher (Deacon)
on Jul 12, 2012 at 09:05 UTC ( #981338=note: print w/ replies, xml ) Need Help??


in reply to Why PM needs web stats

Both solutions provide a much needed insight into actual user behavior. PerlMonks is a site that recognizes that ‘Content is King’ but how do the users interact with it? How do people find content on PerlMonks? Considering the feedback I have received to drop IE support, how many people are using IE? Android? and which versions? How long do users stay on PerlMonks? What is the average page load time? Are some pages slower than the rest of the site?

I'd question just how useful the information provided by GA would be to a site like this. Yes, GA can tell you what browsers people are using, but if you're designing for particular browsers, you're already Doing It Wrong. That makes sense for some sites, but I think the overwhelming consensus here would be to design for the standards and assume people will use browsers that meet those standards. "Page load time," if it's useful at all, is better gotten other ways. GA can tell you how long the browser takes to get and render the page, but you can find that out for yourself by, you know, browsing the site. To break page load time down to determine whether slowness is caused by network latency, database lookups, etc., you'd have to run something at the server end. If knowing how people find the site (referrals and search keywords) is useful, that can be gotten from the server logs. Ditto "heatmap testing," as far as what menu links people use; the simple menu design means if you want to know how many people click on "Meditations" (for example), just count the hits on it in the logs.

As for using some open-source alternative: we're all programmers here, right? If you want to know what browsers people are using, why not ask the gods if they'll tell you what LogFormat is being used (assuming Apache), and if they would run a simple script you provided to mine that info from the log for a sample period? The same thing could be done to see what menu items are/aren't used, etc.

For the record, I use GA on most of my client sites (they like to look at the graphs), and sometimes it's useful. I also like a lot of "jQuery crap" and think most concerns about Javascript are a decade out of date. I don't worry much about the privacy issue with GA (although my concern in that area is growing), and I doubt it would add any noticeable load time or browser load to the pages. But I think it's questionable how much it helps you to "know your audience," or how useful it is to do so. With my clients, I find that stats like "bounce rate" are frequently confusing and misleading, and tend to cause people to think the stats are more meaningful than they really are.

Aaron B.
Available for small or large Perl jobs; see my home node.


Comment on Re: Why PM needs web stats
Re^2: Why PM needs web stats
by Anonymous Monk on Jul 12, 2012 at 10:38 UTC

    To break page load time down to determine whether slowness is caused by network latency, database lookups, etc., you'd have to run something at the server end.

    Its been said before (tye?) its the database

    If you save-page-as and serve the same page as static page via network, you can see how much faster it loads

    Ditto "heatmap testing," as far as what menu links people use; the simple menu design means if you want to know how many people click on "Meditations" (for example), just count the hits on it in the logs.

    IIRC the top 3 are The Monastery Gates, Newest Nodes, and Recently Active Threads

    The next three are probably "FullPage Chat" (Other CB Clients) and "preview page"

Log In?
Username:
Password:

What's my password?
Create A New User
Node Status?
node history
Node Type: note [id://981338]
help
Chatterbox?
and the web crawler heard nothing...

How do I use this? | Other CB clients
Other Users?
Others lurking in the Monastery: (15)
As of 2014-07-23 18:32 GMT
Sections?
Information?
Find Nodes?
Leftovers?
    Voting Booth?

    My favorite superfluous repetitious redundant duplicative phrase is:









    Results (149 votes), past polls