I need a good count of my Unique Visitors

[webmastertexas]

I know, I know. There are tons of threads on this, but maybe I'm just thick, but I can't seem to find an actual answer in any of them. They all start out well, but then someone pipes in about their way, and someone else disagrees, and the next thing you know there are 50 posts about whose way is better and why the other guy's sucks.

I just want some suggestions, not theory. Please, don't start rambling about this and that and why and not and blah blah blah. I respect all of you, but you people do tend to go on random tangents that just gives me a headache.

Having said that, can you please tell me what YOU use to count your unique visitors, and let's leave it at that. Please, no loooooooong theories. Just tell me what you use. At the end, I'll take the most popular and use that.

Thank you.

[ogletree]

I have sites that the only traffic comes from SE traffic. It is easy to count that I just count the number of searches in my stat program (weblog expert). I know you don't want to hear this but the best stat is the amount of money you make. Just use the same stat program all the time and you can see if you are doing better or worse. If you rank well for any terms your site will be hammered by bots and scrapers that totaly screw up the stats.

[webmastertexas]

The reason I'm asking is because I plan to sell advertising space on my site in 2005, and I'd like to be able to tell potential advertisers all the relevant numbers, such as page views, unique visitors, etc.

[photon]

Many stats program (including the free Webalizer) have metrics for unique visitors. The reason you get 50 thread posts about the best way to count visitors is that server logs were created provide data to manage servers, not web sites. Taking data intended for one purpose, and trying to use it for an entirely different purpose is always going to leave room for lots of errors.

Your best bet is just to choose a stats program, then use it for all of your reporting. It will still be only a "best guess", but at least it will be a consistent best guess.

[mole]

I use awstats. It reports unique visitors, page views & hits.

'nuff said?

[Macro]

Tell them to check it on metricsmarket.com. If nothing's 100% accurate then that is at least easily accessible without providing a password to your stats.

[photon]

From here [support.imhosted.com], about awstats:

Unique Visitors - These are the total number of visits by a unique IP address. This can be a bit misleading because dial-up visitors get a new IP each time they log on so you can have the same person visit different times and give a unique hit.

As I said before, any stats program that relies on server logs is only going to give you a "best guess".

[restricted access]

If you have a log analyzer tool like Analog, that has a configuration file, you can get a very good count of unique IPs minus the bots by simply applying some filters. This won't show bots masquerading as IE or whatever, but it does get pretty close.

Then you have to add in a certain percent that are viewing your site from places like AOL, which caches their own version of the web and serves requests from that, which means the only count you will get is the AOL proxy picking up your pages once. That number can represent 25-30% of your total.

Any unique type visitor stat always has to take factors like this into account, but they do give a decent idea.

[ogletree]

I think se traffic is the best way to advertise. That way you can say you get so many new visitors every day. SE traffic is mostly new visitors. Actulay it really does not matter what you use what matters is what your competiters are using. If they all use the worst unique vistor counter you need to use it to. I'm pretty sure most people advertise unfiltered stats for that number. The higher the better.

[webmastertexas]

Okay, so if unique numbers vary so much depending on what program you use, what IS the best way to come up with a mean number? Say, if I were to succumb to the notion that it is impossible to get a full, accurate number, what IS the best way to achieve a semi-accurate number? On the front pages of my two sites, I use SiteMeter, and it does this very well. But of course it only does it for my FRONT pages, and not the other 5000 pages on my site. Any suggestions how to extend my SiteMeter to the rest of the site? Is that even possible?

[Occupant]

I use StatCounter. It's a free service with some excellent features.

[photon]

...and the next thing you know there are 50 posts about whose way is better...

This is what you're going to start getting into now. :)

[webmastertexas]

I use StatCounter. It's a free service with some excellent features.

Does StatCounter count every visit to your ENTIRE site, or just to the site that you put the StatCounter code on?

[Occupant]

It will count each page that you install the counter code to. StatCounter generates the code for you, so it is just a copy & paste operation. It also can be an invisible counter with a long list of stats for you, including keywords that were used to find your site, the visitors location, the usual operating system,screen resolution/browser info, etc.

[killroy]

Well, the most accurate count would probaly include client side tech such as cookies.

Personally I do the following: I cound sessions, so same guy visiting in two different weeks counts as 2, I use IP+USER_AGENT to differentiate, and I tiem sessions out at 60 minutes. So if two folks brwose form the same IP usign the same user agent but more then 60 minutes apart I correctly detect them as two visitors. If one guy browses, takes a 3 hour break and then browses some more, IO falsly identify that as 2 visits (that's my inaccuracy). In tests I foudn that setting the timeout to 24 hours doesn't cahnge teh visitor count much. This shows me my measure is pretty good, and the inaccuracies introduced from folks taking long breaks in their browsing is minimal.

SN

[ogletree]

What about aol users whose IP changes as they browse.

[Macro]

17 down, 33 to go :) and no clear idea of which one is the most "popular" :)

[Occupant]

StatCounter does use cookies, allows you to be able to set the visit time, and since it does use cookies, no problems with dynamic IP addresses. Besides, they are Webmaster World supporters. No, I have no interest in StatCounter, other than being a satsified user. 18 down, 32 to go.

[nalin]

I use Awstats for log analysis and Conversion Analyst for more in depth visitor stats (does cookie based tracking with an image for noscript tags and seems to handle returning aol users fine).

The former is free.
The latter runs something like $99 a month if under a certain number of hits.

[photon]

since it does use cookies, no problems with dynamic IP addresses

Unless, of course, the users have cookies turned off.

[lambo]

www.xav.com/scripts/axs/ AXS Visitor Tracking, free. It's a terrific program.

[dataguy]

I have a little counter on my home page that says "You are visitor number N,NNN".

Every day when I get up I write down the number and subtract the number from the day before, and then I know.

(19 down and 31 to go.... Sorry, just had to contribute! :)

[webmastertexas]

I have a little counter on my home page that says "You are visitor number N,NNN".

I think people are not paying attention to the original question. I need to know the unique number of visitors for my ENTIRE site, not just my front page. I have that. I use SiteMeter, and it works great, but of course not everyone who visits one of my pages goes to my home page, thus they don't count. I need to count ALL of my unique visitors, and since I have over 1,200 unique content pages, as well as about 3,000 other pages (with pictures), I don't think putting SiteMeter codes on each one would be viable.