SitePoint Sponsor

User Tag List

Page 1 of 2 12 LastLast
Results 1 to 25 of 28
  1. #1
    SitePoint Evangelist hexburner's Avatar
    Join Date
    Jan 2007
    Location
    Belgium
    Posts
    591
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)

    What's really unique about a visitor?

    I don't know exactly how an IP is assigned to a computer -or network- but what I do know, is that an IP is not 'unique'. It changes every now and then and your old IP is assigned to another computer.
    Here in Belgium an ISP called Skynet uses dial-in connections, and for only a year or so they assign the same IP for a longer period of time to a computer or network. But before that, for every time you connected to the internet, you got a new IP. So in a visitor statistics system that visitor is counted more than once.

    Then, imagine I have a home network, one's a PC, the other's a Mac.
    My sister uses the PC, me the Mac. We should both be unique visitors, but we share the same IP and hostname so we're counted as one unique visitor.

    What's your opinion about using the hostname instead of the IP and combining it with information about the operating system -or the browser- the visitor uses by evaluating the user-agent string?
    Last edited by hexburner; Aug 29, 2007 at 05:13. Reason: Bad title
    FOR SALE: 1 set of morals, never used, will sell cheap

  2. #2
    SitePoint Guru MikeBigg's Avatar
    Join Date
    Jun 2004
    Location
    Reading, UK
    Posts
    970
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)
    And consider networls like AOL where the IP address for a given computer can, and does, change for different items on the same page.

    In many organisations, there is a policy to have a common operating environment, ie all workstations are set to work at the same version of software and updates are controlled to ensure this. In addition, it is also likely that many if not all users will access the internet through the same router. So, in theory you cold have many users with the same IP address and user agent accessin a site, especially if it is as a result of following a link in a viral email.

    Step forward the cookie. This, too isn;t fool-proof. However you can use combinations of all this information to come up with a reasonably reliable way of determining unique visitors to a site.

    I would say, that I think the unique visitor number is meaningless without a time period. "1000 uniques" on its own means little. "1000 uniques per hour" is quite useful.

    Mike

  3. #3
    SitePoint Evangelist hexburner's Avatar
    Join Date
    Jan 2007
    Location
    Belgium
    Posts
    591
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)
    So, if someone claims to have 100.000 unique visitors per month, the actual number will be a lot less?

    How does the AOL network work then? Wouldn't changing the IP-address for every computer connected through AOL as ISP cause alot of overhead?
    That can't be right, can it? :s
    If so, wouldn't AOL be the ultimate ISP to use for illegal practices on the internet then?
    The IP changes, but what about the hostname? Does that change?
    Otherwise it would be best to use the hostname and cookies to do this.

    Using cookies ain't fool proof. Firefox clears all my cookies and other private data as soon as I quit Firefox, automatically.
    FOR SALE: 1 set of morals, never used, will sell cheap

  4. #4
    SitePoint Wizard silver trophy kyberfabrikken's Avatar
    Join Date
    Jun 2004
    Location
    Copenhagen, Denmark
    Posts
    6,157
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)
    Quote Originally Posted by hexburner View Post
    How does the AOL network work then? Wouldn't changing the IP-address for every computer connected through AOL as ISP cause alot of overhead?
    The network equipment does that. It's a rather simple 1:1 mapping, so it isn't that expensive in CPU-cycles. The equipment is build for it anyway. Routers do something similar, when mapping the external IP to an internal IP. It's called NAT.

    Quote Originally Posted by hexburner View Post
    If so, wouldn't AOL be the ultimate ISP to use for illegal practices on the internet then?
    Yes, except they log the data, so the police would be able to track down who had which IP at a given time.

    Quote Originally Posted by hexburner View Post
    Using cookies ain't fool proof. Firefox clears all my cookies and other private data as soon as I quit Firefox, automatically.
    Not unless you have some sort of paranoia-plugin or similar. All "normal" users will have cookies enabled, and they can transition between browsing sessions (Eg. they aren't deleted when the browser is closed)

  5. #5
    Programming Since 1978 silver trophybronze trophy felgall's Avatar
    Join Date
    Sep 2005
    Location
    Sydney, NSW, Australia
    Posts
    16,786
    Mentioned
    25 Post(s)
    Tagged
    1 Thread(s)
    Quote Originally Posted by hexburner View Post
    So, if someone claims to have 100.000 unique visitors per month, the actual number will be a lot less?
    OR a lot more if all their visitors are on AOL or company networks sharing one IP between hundreds or thousands of people.
    Stephen J Chapman

    javascriptexample.net, Book Reviews, follow me on Twitter
    HTML Help, CSS Help, JavaScript Help, PHP/mySQL Help, blog
    <input name="html5" type="text" required pattern="^$">

  6. #6
    SitePoint Evangelist hexburner's Avatar
    Join Date
    Jan 2007
    Location
    Belgium
    Posts
    591
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)
    I see. It sounds to me that there are no specific standards to be followed for assigning IP's, or does AOL do this for a reason?

    And, like some developers claim, using an IP-address in a session to do a check-up to validate user access will not work properly for AOL users and alike.

    And how does this reflect on the hostname with AOL users?
    As I recall, IP's change, but hostnames don't, right?
    Or is there a way to use the MAC-address from the network card instead of the hostname, which I doubt?
    FOR SALE: 1 set of morals, never used, will sell cheap

  7. #7
    Grumpy Minimalist
    Join Date
    Jul 2006
    Location
    Ontario, Canada
    Posts
    424
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)
    Quote Originally Posted by kyberfabrikken View Post
    Not unless you have some sort of paranoia-plugin or similar.
    Firefox 2 comes with a built-in "clear private data" feature, although it is disabled by default.

  8. #8
    Programming Since 1978 silver trophybronze trophy felgall's Avatar
    Join Date
    Sep 2005
    Location
    Sydney, NSW, Australia
    Posts
    16,786
    Mentioned
    25 Post(s)
    Tagged
    1 Thread(s)
    Quote Originally Posted by Tarh View Post
    Firefox 2 comes with a built-in "clear private data" feature, although it is disabled by default.
    So do many firewalls and some of them have it on by default.
    Stephen J Chapman

    javascriptexample.net, Book Reviews, follow me on Twitter
    HTML Help, CSS Help, JavaScript Help, PHP/mySQL Help, blog
    <input name="html5" type="text" required pattern="^$">

  9. #9
    SitePoint Zealot twistie's Avatar
    Join Date
    May 2007
    Location
    Tonwsville, Australia
    Posts
    127
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)
    Yes, I wish there had been a way integrated into the IP system to identify individual users by computer or even user account. For example my school has over 700 users all routed through a single connection. But even if 200 people from the school visit a site in one day it is seen as only one unique user.
    "When you say 'I wrote a program that crashed Windows',
    people just stare at you blankly and say 'Hey, I got those
    with the system, for free'." (L. Torvalds)

  10. #10
    SitePoint Author silver trophybronze trophy

    Join Date
    Nov 2004
    Location
    Ankh-Morpork
    Posts
    12,158
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)
    Quote Originally Posted by hexburner View Post
    So, if someone claims to have 100.000 unique visitors per month, the actual number will be a lot less?
    Quote Originally Posted by felgall View Post
    OR a lot more if all their visitors are on AOL or company networks sharing one IP between hundreds or thousands of people.
    Or if you use IP masquerade, so that a company with hundreds of users will appear to be using a single external IP number.
    Birnam wood is come to Dunsinane

  11. #11
    Non-Member
    Join Date
    Jul 2007
    Posts
    101
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)
    All use different terms to describe it. I would say that unique visitor is visitor comes ones per exact period of time.

  12. #12
    SitePoint Evangelist hexburner's Avatar
    Join Date
    Jan 2007
    Location
    Belgium
    Posts
    591
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)
    Ok, I understand that, but is there a solution for this problem?
    How can you measure statistics correctly then?
    FOR SALE: 1 set of morals, never used, will sell cheap

  13. #13
    SitePoint Guru MikeBigg's Avatar
    Join Date
    Jun 2004
    Location
    Reading, UK
    Posts
    970
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)
    In addition to the regular IP address, some server requests will include a couple of other fields which can be useful ... HTTP_X_FORWARDED_FOR and
    HTTP_CLIENT_IP. The first may actually be a list of IP addresses separated by commas. It should be said that some routers add the data, some don't and some will remove it!

    As for measuring statistics correctly - I don't think it is possible to *exactly* count the number of unique visitors for a given period.

    You can use an existing system, like analog stats or webalyzer or google analytics or Dan Grossman's w3counter. Chose one that you like and use that.

    I would guess that each of the systems mentioned have an algorithm that was thought about and that can be justified.

    Or you can develop your own algorithm and generate the stats yourself at least then you know exactly what is being measured.

  14. #14
    SitePoint Evangelist hexburner's Avatar
    Join Date
    Jan 2007
    Location
    Belgium
    Posts
    591
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)
    It's not really about counting visitors, it's about IP's and hostnames in general.
    From the post above I've learned that IP's change, but no-one mentions the hostname. I can't find thrustworthy information when googling for it, but I get results for changing the IP and the hostname on a particular OS.

    As IP's change, and as far as I know hostnames don't, it's better to use the hostname instead of the IP-address, right?
    Can anyone confirm that hostnames don't change when you get a new IP assigned?
    FOR SALE: 1 set of morals, never used, will sell cheap

  15. #15
    SitePoint Addict
    Join Date
    Sep 2006
    Posts
    368
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)
    hostnames change, ips change

    thats just the way internet is, stateless, live with it

    if people had one ip assigned per person (quite possible with ipv6) it be possible then, there be no spam also on the net as well but there also be no privacy...

    there is no standard for measurement but webmasters understand things such as "my webalizer reports XYZ users" or "the site receives XXXX users according to analytics" or back in the early days page hits

    oh btw AOL is not the only one, i seen connections from singapore for example changing ip on every request, companies do this when they are short on ips and other reasons such as filtering all traffic at the edge of their networks

    its easy to track registered users tho, they have to login to use functionality in most sites nowadays

  16. #16
    SitePoint Author silver trophybronze trophy
    wwb_99's Avatar
    Join Date
    May 2003
    Location
    Washington, DC
    Posts
    10,625
    Mentioned
    4 Post(s)
    Tagged
    0 Thread(s)
    Big question here is "how long do you want to ID the visitor." Inside a single visit, session cookies are effective, and tracking software can understand them. Using multiple sessions, I tend to cookie visitors with a GUID so you can get a handle on repeat visitors. Or at least repeat visitors from the same computer and user profile.

    But, end of the day, the only real way to be certain of a single user's activity is to put things behind logins and track the patterns of logged-in users. Why do you think Yahoo and Google are so into getting you to create an account and stay logged in? What is your clickstream worth?

    PS: Thread featured.

  17. #17
    SitePoint Enthusiast
    Join Date
    Jan 2003
    Posts
    46
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)
    There really isn't a way to track uniques. Along with all the things mentioned above, what about libraries and other public computers, is the same machine, same ip, same host, but a different person.

    Unless you require your users to login and track their logins and visits to their account, your out of luck.

    Javascript/cookies can be used to track visitors a little more precisely, but then some people turn javascript off so it doesn't work there either...

    Mubs

  18. #18
    Sesame Street Iimitk's Avatar
    Join Date
    Feb 2006
    Posts
    662
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)
    Wow.. could the Web be such a fuzzy place? Are all the millions Google generate through AdSense & AdWords advertising are dependent on a stateless mechanism? I'm really interested in knowing how they've addressed such difficulties in their business models.
    Imagination is more important than knowledge. - Einstein

  19. #19
    SitePoint Addict
    Join Date
    Sep 2006
    Posts
    368
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)
    I'm really interested in knowing how they've addressed such difficulties in their business models.
    by collecting humongous amounts of information thru adsense/adwords/search/analytics/gmail and so on

    then data-minining the data to retrieve interesting bits

    they can with a good degree of confidence track an average user, but its not foolproof (hence the google accounts)

    the amount of data google collects is scary

  20. #20
    Sesame Street Iimitk's Avatar
    Join Date
    Feb 2006
    Posts
    662
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)
    Here's an interesting former thread:
    http://www.sitepoint.com/forums/showthread.php?t=475616

    Led by the interest in the topic here, I made few deeper web searches & found this interesting guide:
    http://www.webanalyticsassociation.o...I-20070816.pdf
    Imagination is more important than knowledge. - Einstein

  21. #21
    Programming Since 1978 silver trophybronze trophy felgall's Avatar
    Join Date
    Sep 2005
    Location
    Sydney, NSW, Australia
    Posts
    16,786
    Mentioned
    25 Post(s)
    Tagged
    1 Thread(s)
    If you want to track an individual visitor accurately during one browser session then you need to establish a session when they arrive on your page. That session will then serve to identify them throughout their visit.

    If you want to track them accurately across multiple visits then you need to ask them to login when you establish the session so that you can match the current session to their past visits.

    Any other method will not positively identify a particular visitor as being the same person on different visits.

    The way the internet was originally designed makes it just about impossible to track a visitor accurately unless they let you do so.
    Stephen J Chapman

    javascriptexample.net, Book Reviews, follow me on Twitter
    HTML Help, CSS Help, JavaScript Help, PHP/mySQL Help, blog
    <input name="html5" type="text" required pattern="^$">

  22. #22
    SitePoint Wizard
    Join Date
    Jul 2003
    Location
    Kent
    Posts
    1,921
    Mentioned
    1 Post(s)
    Tagged
    0 Thread(s)
    So someone could come back with a different IP address. Big deal.
    It is likely that someone else could be assigned that IP address and NOT be counted as unique. So when you deal in 10s of thousands of visitors, does it matter if it really only was 90&#37; of that number, or 80%? I don't think so to be honest.

    And if someone comes back every day, with a different IP address, at least they are coming back, so again, big deal if it changes, you have a regular visitor!

    And if the number of unique visitors goes up over time, again it doesn't matter, it's going up, either as really uniques increase or regulars come back. Both are a good thing.

  23. #23
    Follow Me On Twitter: @djg gold trophysilver trophybronze trophy Dan Grossman's Avatar
    Join Date
    Aug 2000
    Location
    Philadephia, PA
    Posts
    20,580
    Mentioned
    1 Post(s)
    Tagged
    0 Thread(s)
    User agent strings vary between computers more than you'd think. I've recorded tens of thousands of different user agent strings for Windows XP computers running Internet Explorer 6.

    W3Counter identifies unique visitors by a combination of IP address, sessions, cookies, and user agent strings. It's able to differentiate unique computers behind a single IP address with pretty good accuracy.

  24. #24
    Programming Since 1978 silver trophybronze trophy felgall's Avatar
    Join Date
    Sep 2005
    Location
    Sydney, NSW, Australia
    Posts
    16,786
    Mentioned
    25 Post(s)
    Tagged
    1 Thread(s)
    Both IE and Firefox allow you to change the useragent to anything at all so there are infinite possibilities.

    Opera and Safari are more limited in the values that they use.

    Opera will automatically change its useragent to identify itself as IE or Firefox to some of those web sites that stupidly check the useragent to decide which browsers they will allow in. The file that controls this can be edited by each individual but is also updated automatically by Opera as well.

    Statistically, a large enough sample will give fairly accurate numbers on how many different visitors you are getting but it is just about impossible to identify the same visitor on multiple visits unless they want to tell you who they are and there are lots of different ways that they can conceal who they are such as using different proxy servers for each visit, changing their useragent at intervals etc. It is only that the number of people actively hiding their identity is so small that the statistics are as accurate as they are. If lots of people started actively hiding who they are on the web then all the statistics would become meaningless.
    Stephen J Chapman

    javascriptexample.net, Book Reviews, follow me on Twitter
    HTML Help, CSS Help, JavaScript Help, PHP/mySQL Help, blog
    <input name="html5" type="text" required pattern="^$">

  25. #25
    SitePoint Member
    Join Date
    Jun 2007
    Posts
    3
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)
    Thanks for the Info Friends


Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •