New look homepage for the site

Regular visitors here will notice the new look front page which, for now, contains lists of the most recent and most viewed articles, as well as a list of the top referring sites who send visitors over here. The stats are calculated on request and generally cover the preceding 7 days. The CMS collects various snippets in its logs, to complement the existing Apache log information and I thought it would be a good idea to make better use of the information gathered.

It is as accurate as the visitors allow it to be, with regard the referers and obviously can't account for those who are so paranoid as to block that HTTP header when browsing. There is an admin option to filter out particular domains in the referer calculations, in case unscrupulous webmasters try to abuse the feature. I'll be surprised if it really needs to be used though, as referers are calculated based on the number of unique visitors that they send here.

Maybe some more stats to follow

Depending on how things go, I may add to the stat attack, with comment related stats, but I'm not doing that as yet, considering that no comments have been posted. If and when they do start to come in, I'll look at the levels and decide whether it will be worthwhile. I may also tinker with the time of measurement, which is currently 7 days for article views and referers. I think that's probably the right length of time to measure these statistics over. Longer will mean more work involved in the calculations and may detract from the statistical relevance. Things change daily on the web and so more than 7 days might make the whole thing too stale.

I've tried to make the algorithms as efficient as possible, but as I've mentioned elsewhere on the site, things aren't exactly optimal with my server at the moment. When the database is being served by the Poweredge 1400SC with its nice fast SCSI drives, I'll have a better feeling about these stat generating functions, but developing it in a slower environment is probably better, as it forces me to keep it as efficient as possible. If the overhead becomes too much, I'll move things around a little and have the stats generated and the results stored for display, on an hourly basis. Fingers crossed, there won't be any need for that and I'm keen to keep cron'd jobs to a minimum, with a couple of scheduled tasks already in place on a daily basis.

Search engine query stats added

I've implemented another statistics function that scans the logs for referrals from the major search engines and analyses the search queries used. The top ten complete search queries and then a breakdown of the top ten individual words used in search engine queries are now displayed. The function does both in one, to keep queries to the site's database minimal and is as efficient as I can make it, but might have some scope for optimisation. Although I've always been comfortable with optimising sites for best search engine visibility, I'm being drawn into it as a topic of interest in its own right.

The analysis is safe and accurate, although I do see how it might be manipulated I want to see if and how that happens, before I design around it. For now, it is enough to make things safe from a security standpoint and deal with tackling any content manipulation, as, if and when it occurs. Analysis of keywords and search queries are going to be a big part of this CMS, according to the plans I have for various features and so the search engine analysis is just one initial step in that direction. Future steps should stay fairly behind the scenes and any visual evidence of it should be pretty subtle. I can see the way that content management systems are going, looking around at the various blogging systems available, and I want to put together my own vision of how it should be done, whilst still providing the features that seem to be in demand.

Comments to this article

Post a comment:

Like this page? | |