I track over 1000 sites with analytics and its completely wrong all the time. I have done numerous test with iframes from one site to another and the results for one site dont match the other. The stats are just random, they are similar to doing a backlink count on google, they only show what they feel like for that day
Why does Google Analytics report different values than some other web analytics solutions? - Here's the answer ".../support/googleanalytics/bin/answer.py?hl=en&answer=55614" (sorry Im not allowed to post a link. Plz replace ... with google.com) GG Analytics is quite enough for me now. Nothing to complain but the referrers' specific IDs or detailed URL
The whole concept of accuracy is irrelevant no matter what analytics package you use. Analytics cannot accurately count real visitors, it never could, and it probably never will. The figures are all wrong, we just don't know how wrong. But it doesn't mater. It's trends you should be measuring. As long as you're consistent (use the same analytics package to measure the same metrics), your analytics package should provide valuable data. Forget the numbers, just look at the trends. Are things improving or not? Traffic going up or down, at what rate, etc ?
I have seen the same with analytics and Urchin. Urchin is always about 3 times higher then analytics. If awStats use setting like session, then it would give misleading count. A user can have multiple sessions on same visit. Again if IP address is used, users with dynamic IP would be counted as multiple.
Analytics can only track when the javascript or if js disabled (the image it shows) is executed. A bot that scrapes your site will not load the image, nor urchin.js. AWStats is server side and logs all server activity, which means that your AWStats is showing bot activity as well as users who have disabled or blocked urchin.js from running (or have javascript + images disabled). AWStats can detect some bots and will show you that in stats, while Analytics will only show users and (bots) which execute the js/images, it's more likely that they are users since only a really badly programmed bot would execute javascript or images on a site. Analytics is a good measure, especially if you're using Adsense because those who block Adsense most likely also block Analytics - so you get a better number of the users that matter (those that actually see your ads). Of course this is all heresy since I have no facts to back it up.
I use SmarterStats which uses my IIS logs to track visitors, including bots and real users. It shows me at about 1,000 unique visitors per day while Analytics is about 200 per day. Analytics is at least reliable enough to follow my real visitor count by about 20%. If SmarterStats shows 2,000 visitors, Analytics shows 400. I'm not sure why the Google Analytics tool is so far off, or what algorithm it uses to define a "real" visitor, but like was mentioned upthread, as long as you are using the same tool to measure your traffic, you can get a good gauge of trends. (hopefully you're trending upwards!)
Maybe SmarterStats is showing pageviews instead of uniques. Analytics is pretty accurate on the stats. It matchs up with awstats and the other third party stats program I use.
You've made some great points. There are also a couple other factors. Server lag. Depending on the popularity of your site, analytics will begin to dog and stop reporting even the hits that come in and should be seen (client is not blocking and JS is turned on). I did some research on this for quite awhile. I have a script site that i track. I've done this tracking both with Analytics and a highly customized installation of AWStats. I used plugins which gathered the same data as analytics which would normally be the part that sets it above the rest. eg: tracking conversions and other adsense items. Over a 3 month period that amount of missed hits from analytics was staggering. I also tracked stats using the same installation of AWstats and Google analytics. This site got somewhat average traffic. The accuracy was much closer this time. Which leads me to believe Analytics will at some point start missing your incoming data. My guess would be this is a built in feature to stabilize things and keep from backing the whole system up from a few hundred BIG sites. The results I got were pretty close to what I expected. Google Analytics has a very good place in the tracking market. But you need to know what to expect. Use it for watching trends, and maybe conversion tracking. but do NOT use it for accuracy of traffic. It's not accurate. Even if you were to remove the javascript being turned off I think you'd find it's off. I was fortunate enough to be able to track this on 3 sites because I have 3 very different sites. 1. is a blog. a personal one. I get low traffic. around 120,000 page views per mmonth. Accuracy here was acceptable, but still off. 2. is a blog/news site. Currently getting a little over 500,000 page views per month. 3. Is a script site I wrote which has grown popular over the last year. Currently it's getting a little over 39,000,000 page views per month. This allowed me to scrutinize a much larger chunk of data. Site 3 was actually over 20% off. While sites 1 and 2 were closer to 8 to 10 %
I think that Analytics is better because it tells you loads but the problem is that it only counts if the whole page is loaded while the server-side things like webalizer and awstats don't.
I have heard many said the analytics is not always accurate. Anyway, i still using analytics till now because it's the best amongst the free.
Thanks for all the comments guys, some nice opinions here. I really like TangoUK's comments about the trends being the important bit, which is true for the most part, on the other hand though its nice to know how many visitors you're getting too, when trying to attract advertisers etc, and work out if you're performing decently with generating leads/sales against the number of visitors you're getting. I'll look into some of these other systems that have been mentioned and try and get a third reading on the stats, then see if I can draw a conclusion on that. Thanks all. Matt