Counting click-throughs

Petras has asked for the wisdom of the Perl Monks concerning the following question:

Replies are listed 'Best First'.
Re: Counting click-throughs by Aristotle (Chancellor) on Nov 12, 2003 at 05:42 UTC
Well, not that this has anything to do with Perl, but the usual method for doing this is to use a script on the server that redirects to the link target. Instead of sending people to http://example.org, you send them to http://example.com/link.cgi?url=http://example.org where the link.cgi takes note of the click and then sends a redirect to the given target. Makeshifts last the longest.	[reply]
Re: Counting click-throughs by barrd (Canon) on Nov 12, 2003 at 14:43 UTC
Hiya Petras, Thought I'd carry on a little from where Brer Aristotle left off. I've done many types of this over the years and would like to give a wee bit of advice. I started out by using flatfiles and the exact same technique as above to keep a count (the link went through a CGI opened the appropriate text file and then auto increment the number). However this gets messy pretty quickly and is hard to maintain. Step in relational Db. I use MySQL but any other SQL Db would suffice. the reason I do this is to allow better control via a backend admin script to control all the elements of the promotion. Below is a basic typical setup I might have for an SQL table containing the fields; UID (Unique ID) Name of promotion (or in your case charity) Clicks (number of hits) Image (usually just a name or path) URL link Header text - optional (HTML) Footer text - optional (HTML) Created (date promotion started) LastMod (date link was last clicked in this case) Active ("Yes" or "No") This means that your front end (HTML) page interacts directly with your back end admin script (via the Db) meaning that if you are capable of doing this (dynamic page building) not have to edit any hard coded HTML. I gave that up a "long time ago"... ;) Anyway hopefully that might have given you an insight to a more proactive way of handling this situation. If you have any other questions either reply to this post or `/msg barrd` in the ChatterBox. Best of luck...	[reply] [d/l]
Re: Counting click-throughs by Theo (Priest) on Nov 12, 2003 at 20:25 UTC
Check with his ISP. My ISP keeps stats. One set is call "Exit Pages", showing what page someone was viewing when they left my site. Sounds like exactly the thing you're looking for. Hmmm, maybe not. OK, try this: assuming you do have those kinds of stats available, have each link of interest point to a page that just does an immediate transfer to the site of interest. That page would be logged as the exit page and if appropriatly named, would tell him where the visitor was going. The refresh meta tag looks like this: `<META HTTP-EQUIV="refresh" CONTENT="0; URL=http://www.example.com">` [download] (This does seem like a clumsy work-around but I think it's workable) -theo- (so many nodes and so little time ... ) Note: All opinions are untested, unless otherwise stated	[reply] [d/l]
Re: Counting click-throughs by chanio (Priest) on Nov 13, 2003 at 06:02 UTC
I was thinking of doing something similar to this. There is a more generic way, I think. Using the Redirect function of modperl. It is said that modperl works quicker that a normal script. So, I think that in your script you might have more time to check that the address of the redirect was one that you where wanting to count. Those interesting addresses might be in a DB and if they are not there, then you don't need to count those redirections...	[reply]
(z) Re^2: Counting click-throughs by zigdon (Deacon) on Nov 13, 2003 at 16:12 UTC
Time doesn't have to be an issue - when I write something like this, the script that is called to count does the redirect as the first thing, then closes STDOUT. At that point, the user is already redirected, and your script can take all the time it needs to analyse and record the click. -- zigdon	[reply]