Top 50 pages instead of 100 log-size

[fred2028]

In the popular pages section, it only shows you the top pages visited by the last 100 people. Why not make it so it only shows you, say the top 50 pages visitors look at instead? This way, there is less need for log size, and so it will also benefit webmasters more since they will see which pages are most popular over the entire life of their website, not just over the last 100 visits. Also, for this popular pages thing, it would be a good idea to remove the homepage of the user's site, since the homepage will obviously be most popular.

[webado]

Well, all stats are only based on the log content, which is the latest 100 pageloads at all times. This is the only way to ensure no more than 100 log entries are used up for any one account. The Popular Pages statistic simply means that all the pages visted in those latest 100 pageloads get sorted in decreasing order of how often they each got visited within those same 100 latest pageloads - so it has nothing to do with how many visitors viewed them, only with how often they got loaded. This statistic will never contain more than 100 different page url's, and will be limited by the total number of pages a site has anyway with he proviso that www and non www url's are treated separately at present.

The home page itself isn't so easy to determine as you may think. For most sites it may well be index.html, or index.htm, or index.shtml, or index.php or index.asp or any file called index with any other suffix, or it may be even called somehting entirely different according to other conventions such as maybe home.html and variations thereof. There may not even be any particular homepage tracked either, since tracking depends entierly on whatever page of the site the user decided to track by a particular project code. Firthermore when one access the domain, the homepage file name may not even appear, if it's the default for the server, or the site is framed.

[oolong]

It's not quite that all stats are only based on the log content, is it? The Summary stats go much further back than the last 100, and illustrate a general principle: It's not always necessary to store the full details of logs in order to accumulate useful statistics.

For example, it would (as far as I can see) not add a great deal of overhead if you were to allow users to nominate, say, 10 pages which they want to track numbers visiting, and 10 pages they want to track visitors coming from, over time. That'd be a great feature, too. I'd really love to know just what percentage of my users come from StumbleUpon (http://www.stumbleupon.com) over the course of a few weeks, for example.

Incidentally - on the 'home page' thing - it's actually not at all obvious that the 'home page' will be the most visited page of every site. Mine generally ranks about fourth, with far more visitors going straight to Resonata (http://fergusmurray.members.beeb.net/resonata.htm), Trigonometry (http://fergusmurray.members.beeb.net/trig.htm) or Trochor (http://fergusmurray.members.beeb.net/Trochor.htm).

[webado]

Well, the summary is nothing but an accumulator of values, with no details, for each day. As each hit is recorded the summary entry for that day is updated with what has been determined from that hit.

For sure the home page is actually rather irrelevant on its own. It might never or rarely get visited as you say. As far as the project is concerned, it's just another url, it doesn't even have to be on the same website. The counter counts regardless of what page and what website it is placed, including on one's own pc.

Perhaps it will be considered for implementation, as long as it a finite set of pieces of information, nothing variable. Nominating specific pages for stats may be feasible. But keeping a running statistic of your most popular page for instance may not be, because that means that any and all pages ever visited which contain that counter will have to have their own pigeonhole in order to accumulate the hits. This is open-ended, thus not actually feasible under the present configuration, and certainly not for the basic service I should think.

[oolong]

Exactly my point - I'd love to have a simple accumulator of values for each of a small number of referrers, and a small number of my sub-pages, as well as for each day.

What do you think? Easy enough?

Cheers!

[webado]

Sure, easy enough .... especially since I'm not the one to do it :lol:

Just wait and see what special goodies Aodhan has reserved for the Advanced Statcounter Project and maybe a lot of wishes will be fulfilled. :wink:

[robinev]

Exactly my point - I'd love to have a simple accumulator of values for each of a small number of referrers, and a small number of my sub-pages, as well as for each day.

Yeah. Easy enough in one way, but not so easy to pay for the massive increase in database space and server load required to keep that kind of expanded info for the tens of thousands of sites tracked by StatCounter for free.

There have been plenty of hints that the Advanced StatCounter, when it finally arrives, may offer those kinds of features and more, but I think we can expect to be charged for it if we require that kind of more complex reporting.

[oolong]

Hi robinev,

It was mostly from the point of view of database space that I was looking at this as 'easy enough'. As far as I can see, we currently have 100 entries in a database (in the free service) each containing two URLs, an exact date and time, and an IP address (I'm not sure if all the additional information listed is simply extracted from the IP address, or whether that's stored separately). On top of that, we have a seemingly ever-increasing number of entries for days, each storing four numeric values (Page Loads, Unique Visitors, First Time Visitors, Returning Visitors).

So ten more entries (say), each storing just a URL and a number, surely amounts to a very small portion of the data already being stored...?

I don't know how great the increase in server load would be - each referral would need to be checked against the ten chosen entries, I guess; it's possible that that would add up to a substantial increase - anyone able to estimate how much processing would be involved, relative to how much is already performed?

[webado]

In order to establish which are the X most popular pages you must have log space for all different pages that have your counter code and have ever been visited. That's the problem, this is an undefiend space requirement, it can grow to any size, out of control.

[oolong]

Thanks Chris,

That's a fair point - and it's why I suggested that users be required to nominate a few pages to track, rather than trying to generate them automatically (which, as you say, is pretty well unfeasible). Ideally, they could use wildcards to catch referrals from more-or-less the same place.

I've got a pretty good idea who my top few referrers are, but I'd still like to know how they weigh up against each other.

Make sense?