better stats - javascript or server-side?

[brianzajac]

Ok, so we all want to measure traffic "correctly". But, we also want to have 'good stats' for our clients. Through my research, I've found some interesting facts on stats:

Javascript - Adding special javascript code to every page on your site. You can then view stats on the supporting website.
The good: More accurate stats and, in some cases, are free (i.e. sitemeter.com offers a great system).
The bad: Doesn't count all statistics (like web bots and other miscellaneous incoming traffic) & uses a third-party's collection of statistical data (i.e. they can use your data for other things)

Server-side - Installation of a statistics system on your server. The software then analyses your log entries and disseminates the data for your viewing pleasure.
The good: Counts everything. Great for showing 'more traffic' for your web clients.
The bad: Counts everything...not as accurate. Plus, if you own the server, you've got more responsibility to deal with. Lastly, server stats (really good ones) can get quite expensive.

So, which do you find more effective for your marketing campaigns - stats or javascript?

[ADAM Web Design]

Server-side with a proper analyzer for the raw logs (Live Stats, Webalizer). They'll know which ones the bots actually are for the most part and will be a lot more likely to be accepted by the client than a Javascript.

Also, putting code on the site for a hit counter looks tacky and who else really needs to know how many visitors a site has other than the site owner?

[brianzajac]

ADAMWebDesign,

Take a look at http://www.susl.com/index.asp See any counter? Now, highlight the area underneath " ©2005, SUSL Soccer League, All Rights Reserved". You'll see it then & can access the stats. Keep in mind this is a freebie version...other paid versions allow you to make it invisible & unreachable.

Also, I've been using Live Stats through an ISP for awhile and noticed that the traffic trend is MUCH higher than the javascript. Think the javascript is incorrect? I used multiple ones and came up with the same result. My guess is that Live Stats is counting the web bot traffic.

[Spectur]

Go with ser stats . they are more accurate andy day of the week..

If you parse the raw logs you can drop the webbots and get a good count.. you can also see if your site is get dos attacked or what not...

Just my 2 cents worth..

[brian.mark]

Because of some of the weaknesses of each, we use a combination of both.

JavaScript (page tagging) stats can show you a "Truer" path for visitors in near real-time. For example, clicking the back button doesn't show up in log files, but JavaScript stats will show that the prior page was reloaded.

There are many decent free stats packages available for analyzing log files. They're not that hard to write, either. I use a combination of Webalizer and a custom written stats package (basically just validating some numbers against Webalizer).

For page-tagging, we use IndexTools. We were able to do a funnel analysis and get some pretty cool stats about people other than our employees that browse the website. Being able to do that in real-time also allows us to see who is browsing during peak times. We've located a few bots roaming our site from competitors this way. This has been primarily form Rivalwatch, but there have been a few others as well. Some firewall rules and it is easily corrected.

Since each has strengths and weaknesses, I feel it is a good idea to use both whenever possible.

Brian.

[ADAM Web Design]

The Live Stats would naturally be higher, due to the simple fact that Javascript can be disabled.

Also, the fact that you've managed to use an invisible image to hyperlink to your site meter stats, IMHO, is only a slightly different kind of tacky. A user hovering his/her mouse might find it by accident (it does happen) and lead him/herself away from your site.

Users don't need to see this stuff. And why direct someone away from your site and confuse them like that?

With regard to the Live Stats comments: it does track web bot traffic, and it is included in the Visitor Info --> Web Site Crawlers section. Whether or not it's included in the Page Views, I can't say for sure, but it does track and separate them.

I do know there's a way to disable specific IPs from showing up in Live Stats, but I can't remember what it is.

So I stand by my original comment.

[timmathews.com]

Go with both.
This way you have the best (and worst) of both. You can compare the 2.
If you can only have/use one, go with server side reporting via webtrends or similar.

[brianzajac]

Since each has strengths and weaknesses, I feel it is a good idea to use both whenever possible.


This statement couldn't be anymore truthful. But let's look at this another way - hosting. On one hand, you've got the server-side stats that show more info & it's hidden...but it comes at an expensive price for any company wishing to do hosting - both for cost & development upkeep (i.e. removing log files, general problems w/ accounts that arise, ect).

On the other hand, you've got javascript - you can find these for free (or for a minimum cost to remove stat counters) and other companies can take the burden off your shoulders should you decide to host for yourself. The info is good, but isn't usually as complete as server side.

So, IMHO, if you're starting to host accounts, I would consider opting for the javascript due to it's less headaches and cost-factor...have someone else deal with the problem. However, as you gain more clientele (and client requests) for a more 'robust' statistical solution, consider the server-side...and have you clients pay the extra. Isn't development time worth that?

Lastly, if you aren't hosting your accounts, I would consider this statement left by an earlier post:


I do know there's a way to disable specific IPs from showing up in Live Stats, but I can't remember what it is.


And neither will your hosting company unless many requests come in for that. You see, setting these minor things up is a custom job...usually made for people who have a dedicated server. But, with all those server-side statistic headaches, it's a time & money issue. So balance (wants vs. needs) and growth are the key to it all...choose wisely.