IP Extravaganza - Help!

[Channel5]

Hi,

I wrote a script that tracks how many page impressions are made. To do this, I take the visitor's IP address and put it into a table with the number of impressions.

The problem is that it's recording way more impressions than I get visitors. When I look at my MySQL table I notice IP addresses that are very similar but different. Example: 205.111.90.14 and 205.111.90.45. I assume they're the same visitor, so it's counting a single visitor twice... sometimes way more.

Does anyone know why one visitor would have several IPs? And what I can do about it, otherwise the counter's completely inaccurate.

Much appreciated

[etully]

There are lots of reasons why someone would have multiple IP's. I don't know all of the reasons but a complete list hardly matters at this point - you just need a solution.

Firewalls, proxies, load balancers, and creative network architectures come to mind as some of the most likely causes.

Check to see if they have a cookie called "IP" set.
If yes, use that as their IP address and proceed as you are currently recording impressions.
If no, then find their IP address from the env variables (as you are currently doing).
Create a cookie called IP and set it's value to the IP you found in the env variables.
Use this IP value to record your impressions.

In other words, their first visit determines their IP. On all subsequent visits, they use the IP from their first visit.

[bradgrafelman]

Tracking users by IP address is a futile effort at best. As etully has already mentioned, there are a variety of reasons why a user's IP address might change - even from on request to another (AOL users come to mind).

On the flip side, there are also places like schools, corporations, offices, etc. where hundreds or even thousands of users will share a single external IP address.

[etully]

Brad is right. IP's are simply the wrong way to do this. So I'm going to modify my answer slightly:

Use a $_SESSION variable or create a unique_id and assign it to a cookie. This will identify a unique user when they return multiple times from various IP's. Now, use that id to record views to each page. This will as close to perfect as you're going to get. (You can't achieve perfection because sometimes the user will see a page from their cache or a firewall's cache and your server won't know about the view and won't be able to record it. Of course, that could be solved with a little Ajax call).

[Channel5]

Thanks for the input guys.

The thing is the solution you gave was how I first coded it, using a session and cookie to make the count. When the results came back poor, I decided to try the IP approach.

My problem is the count comes back way too high. Comparing it to Google Analytics, it's sometimes over double!

I don't care if it's not 100% accurate. Just as close as I can get it... and, well, double's not close enough.

Do you guys have any other suggestions?

Here's a sample of my current code (I've done the same thing using cookies instead of sessions too - same results):

if(isset($_SESSION['userid'])){
     $userid = $_SESSION['userid'];
}
else{
     $userid = rand(1000, 10000000);
     $_SESSION['userid'] = $userid;
}

$date = date('Y-m-d');

$counters = mysql_query("SELECT * FROM counters
WHERE userid='$userid'
AND impressions='1'
AND date='$date'")
or die(mysql_error());

if(!mysql_num_rows($counters)){
     mysql_query("INSERT INTO counters
     (userid, impressions, date)
     VALUES('$userid', '1', current_date)")
     or die(mysql_error());
}

Something tells me that the problem is that sometimes a visitor is somehow getting more than one userid assigned, and therefore counting more than 1 impression.

Any idea about that or how to solve it?

[dnszero]

Um...Stay away from using random numbers like that. It's basically, randomly, counting unique visitors. Maybe they're unique, maybe they're, spin the Roulette wheel to find out! 🙂

Back to the point: A better solution might be to use the auto-increment ID as both the row key and the $userid.

On the other hand, if you just want to count visitors in a given period of time, they really don't need ID's at all...
1. On each page load, check for a session cookie that you set with the current timestamp
2. if it's there, you know you already counted that visitor (and how long ago)
3. If it's not there (or if its too old): Set the cookie with timestamp, and add 1 to your visitor count for the day (or whatever time period).
4. Of course, this depends on the users having cookies enabled and your desired tracking time period being less than the cookie lifetime.

The requisite link to more info (which really just mean more work):
http://www.php.net/manual/en/book.session.php

My REAL advice would be to use a 3rd party app designed for this, like Google Analytics, or better yet GetClicky, but that wasn't your question. ;-)

[dalecosp]

Statistics are very difficult to make "true". Google Analytics, as you've noticed, is "stingy" with its numbers. There are reasons for this: it ignores all bot visits, plus anything else you tell it to (we, for example, tell it to ignore anything from our network, since that would influence our stats wrongly [refreshing to check our work, for example, and demoing the site to potential customers]).

The other thing about GA is that it uses cookies only, so anyone in incognito mode, or using a "cookie cleaner" or other privacy software, and, perhaps, corporate users with certain security set-ups, aren't counted.

The other extreme end of the stats spectrum is Webalizer and similar tools that do logfile analysis. My boss looks at these as truth, but they do count bots and our office loads and a lot of other stuff that can make things look a tad better than they are in our case.

I'm usually tempted to take numbers from both, chart them and draw a line in the middle to make a claim about our site stats. I also look at particular fields returned by webalizer and do some math to get a more realistic number ... for example, about the only page that's really a page on our main site is index.php, so I'll take the hit count on that page and reduce it by the percentages that show up as Google, Bing, Yahoo, Yandex, and our office network.

HTH,

[feargegolla]

Symbicort is a fixed dose combination of budesonide elimite instruction zithromax wellbutrin Jessica Mantel, Co-Director of the Health Law Azadirachta indica or Neem can be a Each dose should be taken at the scheduled Wouldn’t such a travesty of journalism The Best Handcrafted Wire Findings: Techniques and Kids love 3d characters. There are