Loading ...
Sorry, an error occurred while loading the content.
 

Re: [webanalytics] Re: page tags vs web server logs diagram

Expand Messages
  • Steve
    ... That s not ... quite... true. If you look at the raw HTTP headers sent via a proxy server, you can sometimes see the full chain of proxies a client browser
    Message 1 of 25 , Oct 31, 2007
      On 11/1/07, Paul Holstein <paul@...> wrote:
      > There is only one IP
      > address sent to the web server and the client's Proxy servers will
      > show the same one, no matter where the connection originated.

      That's not ... quite... true.

      If you look at the raw HTTP headers sent via a proxy server, you can
      sometimes see the full chain of proxies a client browser has to go
      through.
      I've never bothered chasing this in detail, but I suspect it's due to
      deliberate proxy configuration (ie set a proxy in the browser) vs
      transparent proxying.

      Unless you configure your server accordingly, only your application
      (eg PHP) will be able to log the proxy address(s) used.

      It's been as issue for me/us on a separate website where we were doing
      filtering against IP Addresses within the PHP application. Was also
      useful when doing investigations when some not_quite_clever_enough
      horrible person was trying to hide their tracks via several proxies.
      The information was all there in the application logs which could
      easily be (manually) joined with the servers logs. A little whois and
      other public information to join and you could pin point the actual
      person with a very high degree of accuracy.


      But all this raises the question: Where DO page tagging solutions get
      the IP address from? Surely NOT the browser or else we'd all be
      logging a lot of traffic from 192.X addresses!
      I'd suggest it's done in *exactly* the same way as stock standard
      server logging. From the image request made of the Page Tag Providers
      Servers. :-)


      HTH?

      Cheers!
      - Steve
    • Steve
      ... Grin. It s complicated because YOU and YOUR issues aren t the problems that the builders of those networks are trying to solve. To be brutally frank: how
      Message 2 of 25 , Nov 1, 2007
        On 11/1/07, Rachel <remtheory@...> wrote:
        > Hmm, I thought I posted this already, but still don't see it, so
        > perhaps not. In any case, I have to say, I'm really confused. I
        > thought that because you no longer have the problem of the ISP running
        > interference by going through, say, Virginia, where web files are
        > stored for speed (AOL is apparently notorious for this), you get a
        > clearer picture of where the user is actually coming from. Is this
        > wrong? Please, help me understand! I hate to think I'm giving out
        > wrong information. Why is this so dern complicated?

        Grin.

        It's complicated because YOU and YOUR issues aren't the problems that
        the builders of those networks are trying to solve.
        To be brutally frank: how easy or hard it is for YOU to identify where
        THEIR users are is not even vaguely on their list of priorities.

        eg.
        All our users at work (4500 of 'em) appear to come via a Canberran
        based gateway. Many of them are not in Canberra or the ACT, but are
        scattered all over Australia and even globally. For a generic YOU to
        know where they truly are? Fuggaboudit. We all come from a single
        NAT'd IP Address. You'd never know where they are.

        I've been in a situation where to send emails to my wife 10km across
        town- in the same Federal Department! - our emails would traverse a
        3000km trip.


        Either way. Does it *really* matter? If you get the odd 1 in 10 wrong?
        Is that such a bad thing?


        I've not got a copy of WAD, but I really doubt Eric lied. :-)
        You may find this gives a rough idea of the size of the problem:
        http://xkcd.com/195/
        Gotta love that dirty great big "Various Registrars" :-D

        Cheers!
        - Steve
      • Mark Vozzo
        Following on from the conversation about proxy servers & configuration, I m not sure if I m about to throw a spanner in the works, but what effect (if any) do
        Message 3 of 25 , Nov 1, 2007
          Following on from the conversation about proxy servers &
          configuration, I'm not sure if I'm about to throw a spanner in the
          works, but what effect (if any) do technologies like Akamai
          (www.akamai.com - a Web Application Acceleration technology) have on
          web server logs, Geo-IP and page tags?

          Is anyone using Akamai and if so, has there been any impact on
          WebAnalytics (be it Page Tags or Web server logs).

          Thanks, Mark
        • chonchobar
          Re: geography As someone noted, unless they re using telepathy - and elaborating on Paul s reply below - Google Analytics (or Webtrends, Omniture, etc., etc.)
          Message 4 of 25 , Nov 1, 2007
            Re: geography
            As someone noted, unless they're using telepathy - and elaborating on
            Paul's reply below - Google Analytics (or Webtrends, Omniture, etc.,
            etc.) is till using plain old HTTP log files to compile their reports
            for you. The only difference is the asset that's being requested.
            Chances are, when you were using log files in house, the things you
            were looking for looked like "xyz.html" or "abc.jpg".

            Using Javascript, the WA pixel host gets HTTP requests that contain
            all the data they want (and that you provide) such as pagename, site
            hierarchy or whatever of their supported tracking features you're
            taking advantage of. So in *their* logs, the requests look like:
            "abc.jpg/page=Home/channel=Retail>Store%20Locator/visitor=1234567"

            Up at the front of each log row - in your logs or the vendor's - there
            is still the IP address of "where" the request originated from. That
            could be an ISP, a corporate proxy or an individual machine. The IP
            address maps back to an address in the DNS registries of who the owner
            of that IP address is. AFAIK, that's where most geography related info
            comes from - the same stuff you'll see doing WHOIS or IP lookups.

            At the end of it all, either Google or you are *still* taking the
            visitors IP address from a log file, and possibly using the same
            service that maps IP addresses to a given geographical location. It
            will be the same IP address (distributed proxies like AOL
            notwithstanding) from the same visitor regardless of using tags or
            logs. But Google is also using logs after you request their 'tags'.

            Google no doubt has their own IP-Geography mapping data which may or
            may not be more precise than any other, but the concept is the same:
            What IP addr did the request come from and what does our data say
            about where the registered owner of that IP lives?

            In short - there's no magic in tags about geography resolution.



            --- In webanalytics@yahoogroups.com, "Rachel" <remtheory@...> wrote:
            >
            > Hmm, I thought I posted this already, but still don't see it, so
            > perhaps not. In any case, I have to say, I'm really confused. I
            > thought that because you no longer have the problem of the ISP running
            > interference by going through, say, Virginia, where web files are
            > stored for speed (AOL is apparently notorious for this), you get a
            > clearer picture of where the user is actually coming from. Is this
            > wrong? Please, help me understand! I hate to think I'm giving out
            > wrong information. Why is this so dern complicated?
            >
            > --- In webanalytics@yahoogroups.com, "Paul Holstein" <paul@> wrote:
            > >
            > > Unfortunately, neither log files nor page tags gather geographical
            > > information. Both gather the IP address only. There is only one IP
            > > address sent to the web server and the client's Proxy servers will
            > > show the same one, no matter where the connection originated.
            > >
            > > To get geographical information, either solution must reference a
            > > lookup table that tries to match up IP addresses with locations.
            > > This is an imprecise science that is not 100% accurate.
            > >
            > > A page tagging solution may make it easier to gather geographic
            > > information, but there is no reason to believe it would be more
            > > accurate.
            > >
            > > --Paul
            > >
            > >
            > >
            > > --- In webanalytics@yahoogroups.com, "Rachel" <remtheory@> wrote:
            > > >
            > > > Wow, you guys are great. I knew it wasn't right, but I was
            > > struggling
            > > > to get the picture in my head. I think it's a bit clearer now. For
            > > > what it's worth, I'd like to explain my bias for page tagging. Not
            > > > trying to have a war, but my rationale is below. At this point for
            > > us
            > > > it's not about choosing which method, it's just about giving the
            > > staff
            > > > a brief understanding as to why we chose it. I'd like to make the
            > > > claim of greater accuracy and I thought I could do it based on my
            > > > reading of Web Analytics Demystified. So, here's the rationale...
            > > >
            > > > 1. Web Analytics Demystified laid out one big pro for me, which is
            > > > understanding where, geographically, our visitors come from (esp
            > > since
            > > > we're working on outreach to other countries). It's my understanding
            > > > that web server logs don't accurately count geography due to proxy
            > > > caching. Maybe it's true for page tags, but I didn't get that from
            > > the
            > > > book.
            > > >
            > > > 2. We're a google-friendly nonprofit (in other words, we've got a
            > > > google grant, so the more google we use, the better. :))
            > > >
            > > > 3. In our case, page tagging is a much easier setup than server
            > > logs,
            > > > and certainly cheaper than buying a web analytics product (FWIW at
            > > > former jobs I've used Webtrends - horrible processing experience -
            > > and
            > > > Urchin pre-Google - great product, but still processing large files
            > > > was a PITA).
            > > >
            > > > --- In webanalytics@yahoogroups.com, "Rachel" <remtheory@> wrote:
            > > > >
            > > > > Hey guys, I wanted a visual to illustrate why page tags (like
            > > google
            > > > > analytics) are better than web server logs because you have to go
            > > > > through the ISP to get visitor information instead of right from
            > > the
            > > > > browser (and therefore improving accuracy). First off, is what I
            > > drew
            > > > > even right? Secondly, is it easy to follow? I'm showing this
            > > during a
            > > > > brownbag about web analytics at my org. I added it as a file in
            > > the
            > > > > forum: http://tinyurl.com/2bx98b Thanks! Rachel
            > > > >
            > > >
            > >
            >
          • Paul Holstein
            I wouldn t say he lied. I read a couple of his books and I know Eric personally. I ve never gotten the impression that he s said anything other than what
            Message 5 of 25 , Nov 1, 2007
              I wouldn't say he lied. I read a couple of his books and I know Eric
              personally. I've never gotten the impression that he's said anything
              other than what I've written.

              I'm sure he'll clarify this shortly.

              --Paul


              --- In webanalytics@yahoogroups.com, "Rachel" <remtheory@...> wrote:
              >
              > Really? So Eric Peterson lied? Or did I misread? :(
              >
              > --- In webanalytics@yahoogroups.com, "Paul Holstein" <paul@> wrote:
              > >
              > > Unfortunately, neither log files nor page tags gather
              geographical
              > > information. Both gather the IP address only. There is only one
              IP
              > > address sent to the web server and the client's Proxy servers
              will
              > > show the same one, no matter where the connection originated.
              > >
              > > To get geographical information, either solution must reference a
              > > lookup table that tries to match up IP addresses with locations.
              > > This is an imprecise science that is not 100% accurate.
              > >
              > > A page tagging solution may make it easier to gather geographic
              > > information, but there is no reason to believe it would be more
              > > accurate.
              > >
              > > --Paul
              > >
              > >
              > >
              > > --- In webanalytics@yahoogroups.com, "Rachel" <remtheory@> wrote:
              > > >
              > > > Wow, you guys are great. I knew it wasn't right, but I was
              > > struggling
              > > > to get the picture in my head. I think it's a bit clearer now.
              For
              > > > what it's worth, I'd like to explain my bias for page tagging.
              Not
              > > > trying to have a war, but my rationale is below. At this point
              for
              > > us
              > > > it's not about choosing which method, it's just about giving
              the
              > > staff
              > > > a brief understanding as to why we chose it. I'd like to make
              the
              > > > claim of greater accuracy and I thought I could do it based on
              my
              > > > reading of Web Analytics Demystified. So, here's the
              rationale...
              > > >
              > > > 1. Web Analytics Demystified laid out one big pro for me, which
              is
              > > > understanding where, geographically, our visitors come from
              (esp
              > > since
              > > > we're working on outreach to other countries). It's my
              understanding
              > > > that web server logs don't accurately count geography due to
              proxy
              > > > caching. Maybe it's true for page tags, but I didn't get that
              from
              > > the
              > > > book.
              > > >
              > > > 2. We're a google-friendly nonprofit (in other words, we've got
              a
              > > > google grant, so the more google we use, the better. :))
              > > >
              > > > 3. In our case, page tagging is a much easier setup than server
              > > logs,
              > > > and certainly cheaper than buying a web analytics product (FWIW
              at
              > > > former jobs I've used Webtrends - horrible processing
              experience -
              > > and
              > > > Urchin pre-Google - great product, but still processing large
              files
              > > > was a PITA).
              > > >
              > > > --- In webanalytics@yahoogroups.com, "Rachel" <remtheory@>
              wrote:
              > > > >
              > > > > Hey guys, I wanted a visual to illustrate why page tags
              (like
              > > google
              > > > > analytics) are better than web server logs because you have
              to go
              > > > > through the ISP to get visitor information instead of right
              from
              > > the
              > > > > browser (and therefore improving accuracy). First off, is
              what I
              > > drew
              > > > > even right? Secondly, is it easy to follow? I'm showing this
              > > during a
              > > > > brownbag about web analytics at my org. I added it as a file
              in
              > > the
              > > > > forum: http://tinyurl.com/2bx98b Thanks! Rachel
              > > > >
              > > >
              > >
              >
            • Paul Holstein
              Great question. If you are using Akamai or any other CDN, you are best off with a page tagging solution. This depends almost entirely, however, on your level
              Message 6 of 25 , Nov 1, 2007
                Great question.

                If you are using Akamai or any other CDN, you are best off with a
                page tagging solution. This depends almost entirely, however, on
                your level of integration.

                In our case, we serve only our images and media files through Akamai
                so our .html pages and shopping cart are always served from our host
                server. However, many companies such as Wall-mart serve their text
                files as well and, consequently, may never actually serve a file to a
                customer directly. In that case, they would want a page tag to
                record the address and other analytical information.

                Akamai does offer log files upon request and does report some
                analytical information about your visitors if you need it but I would
                not think that information would satisfy most analysts and it may be
                hard to integrate.

                --Paul

                --- In webanalytics@yahoogroups.com, "Mark Vozzo" <mark.vozzo@...>
                wrote:
                >
                > Following on from the conversation about proxy servers &
                > configuration, I'm not sure if I'm about to throw a spanner in the
                > works, but what effect (if any) do technologies like Akamai
                > (www.akamai.com - a Web Application Acceleration technology) have on
                > web server logs, Geo-IP and page tags?
                >
                > Is anyone using Akamai and if so, has there been any impact on
                > WebAnalytics (be it Page Tags or Web server logs).
                >
                > Thanks, Mark
                >
              • Eric Peterson
                I m gonna go out on a limb here and say you misread. Or, perhaps I didn t explain this as well as I could. But I wouldn t say I lied ;-) If you re-read pages
                Message 7 of 25 , Nov 1, 2007
                  I'm gonna go out on a limb here and say you misread. Or, perhaps I
                  didn't explain this as well as I could. But I wouldn't say I lied ;-)

                  If you re-read pages 24 and 25 in Web Analytics Demystified, you'll
                  notice that the conversation you reference about IP address is
                  presented in the context of using IP as a unique identifier. Yes, I
                  cite the AOL/geography issue as an example of this, and yes, this is
                  presented under the broad header of "Disadvantages of Log Files", but
                  the point was not that log files are worse for geographic lookup.
                  Paul is correct, both tags and logs depend on IP address lookup for
                  geographic resolution.

                  I apologize for the confusion and will endeavor to resolve this point
                  in the upcoming second edition of Web Analytics Demystified.

                  I hope the group is able to resolve your confusion.

                  Eric T. Peterson
                  Web Analytics Demystified, Inc.
                  http://www.webanalyticsdemystified.com


                  --- In webanalytics@yahoogroups.com, "Rachel" <remtheory@...> wrote:
                  >
                  > Really? So Eric Peterson lied? Or did I misread? :(
                  >
                  > --- In webanalytics@yahoogroups.com, "Paul Holstein" <paul@> wrote:
                  > >
                  > > Unfortunately, neither log files nor page tags gather geographical
                  > > information. Both gather the IP address only. There is only one IP
                  > > address sent to the web server and the client's Proxy servers will
                  > > show the same one, no matter where the connection originated.
                  > >
                  > > To get geographical information, either solution must reference a
                  > > lookup table that tries to match up IP addresses with locations.
                  > > This is an imprecise science that is not 100% accurate.
                  > >
                  > > A page tagging solution may make it easier to gather geographic
                  > > information, but there is no reason to believe it would be more
                  > > accurate.
                  > >
                  > > --Paul
                  > >
                  > >
                  > >
                  > > --- In webanalytics@yahoogroups.com, "Rachel" <remtheory@> wrote:
                  > > >
                  > > > Wow, you guys are great. I knew it wasn't right, but I was
                  > > struggling
                  > > > to get the picture in my head. I think it's a bit clearer now. For
                  > > > what it's worth, I'd like to explain my bias for page tagging. Not
                  > > > trying to have a war, but my rationale is below. At this point for
                  > > us
                  > > > it's not about choosing which method, it's just about giving the
                  > > staff
                  > > > a brief understanding as to why we chose it. I'd like to make the
                  > > > claim of greater accuracy and I thought I could do it based on my
                  > > > reading of Web Analytics Demystified. So, here's the rationale...
                  > > >
                  > > > 1. Web Analytics Demystified laid out one big pro for me, which is
                  > > > understanding where, geographically, our visitors come from (esp
                  > > since
                  > > > we're working on outreach to other countries). It's my understanding
                  > > > that web server logs don't accurately count geography due to proxy
                  > > > caching. Maybe it's true for page tags, but I didn't get that from
                  > > the
                  > > > book.
                  > > >
                  > > > 2. We're a google-friendly nonprofit (in other words, we've got a
                  > > > google grant, so the more google we use, the better. :))
                  > > >
                  > > > 3. In our case, page tagging is a much easier setup than server
                  > > logs,
                  > > > and certainly cheaper than buying a web analytics product (FWIW at
                  > > > former jobs I've used Webtrends - horrible processing experience -
                  > > and
                  > > > Urchin pre-Google - great product, but still processing large files
                  > > > was a PITA).
                  > > >
                  > > > --- In webanalytics@yahoogroups.com, "Rachel" <remtheory@> wrote:
                  > > > >
                  > > > > Hey guys, I wanted a visual to illustrate why page tags (like
                  > > google
                  > > > > analytics) are better than web server logs because you have to go
                  > > > > through the ISP to get visitor information instead of right from
                  > > the
                  > > > > browser (and therefore improving accuracy). First off, is what I
                  > > drew
                  > > > > even right? Secondly, is it easy to follow? I'm showing this
                  > > during a
                  > > > > brownbag about web analytics at my org. I added it as a file in
                  > > the
                  > > > > forum: http://tinyurl.com/2bx98b Thanks! Rachel
                  > > > >
                  > > >
                  > >
                  >
                • Rachel
                  Oh wow, I come back from 3 hours of meetings and I see an email from Eric saying I called him a liar and a bunch of new posts on this subject. This is exactly
                  Message 8 of 25 , Nov 1, 2007
                    Oh wow, I come back from 3 hours of meetings and I see an email from
                    Eric saying I called him a liar and a bunch of new posts on this
                    subject. This is exactly why email communication can be so difficult
                    and frustrating.

                    I in *no way* meant that I thought he was a liar. It was sarcasm. But
                    how would he, or anyone on this list, have known that? I am truly
                    sorry for the miscommunication. I love the book and have found it to
                    be the only book on web metrics that I have not only managed to get
                    through (numbers scare me, can you tell?), but have really enjoyed.

                    I'll scurry away in my confusion over page tags and hope for the best.

                    Be well,
                    Rachel

                    --- In webanalytics@yahoogroups.com, "Paul Holstein" <paul@...> wrote:
                    >
                    > I wouldn't say he lied. I read a couple of his books and I know Eric
                    > personally. I've never gotten the impression that he's said anything
                    > other than what I've written.
                    >
                    > I'm sure he'll clarify this shortly.
                    >
                    > --Paul
                    >
                    >
                    > --- In webanalytics@yahoogroups.com, "Rachel" <remtheory@> wrote:
                    > >
                    > > Really? So Eric Peterson lied? Or did I misread? :(
                    > >
                    > > --- In webanalytics@yahoogroups.com, "Paul Holstein" <paul@> wrote:
                    > > >
                    > > > Unfortunately, neither log files nor page tags gather
                    > geographical
                    > > > information. Both gather the IP address only. There is only one
                    > IP
                    > > > address sent to the web server and the client's Proxy servers
                    > will
                    > > > show the same one, no matter where the connection originated.
                    > > >
                    > > > To get geographical information, either solution must reference a
                    > > > lookup table that tries to match up IP addresses with locations.
                    > > > This is an imprecise science that is not 100% accurate.
                    > > >
                    > > > A page tagging solution may make it easier to gather geographic
                    > > > information, but there is no reason to believe it would be more
                    > > > accurate.
                    > > >
                    > > > --Paul
                    > > >
                    > > >
                    > > >
                    > > > --- In webanalytics@yahoogroups.com, "Rachel" <remtheory@> wrote:
                    > > > >
                    > > > > Wow, you guys are great. I knew it wasn't right, but I was
                    > > > struggling
                    > > > > to get the picture in my head. I think it's a bit clearer now.
                    > For
                    > > > > what it's worth, I'd like to explain my bias for page tagging.
                    > Not
                    > > > > trying to have a war, but my rationale is below. At this point
                    > for
                    > > > us
                    > > > > it's not about choosing which method, it's just about giving
                    > the
                    > > > staff
                    > > > > a brief understanding as to why we chose it. I'd like to make
                    > the
                    > > > > claim of greater accuracy and I thought I could do it based on
                    > my
                    > > > > reading of Web Analytics Demystified. So, here's the
                    > rationale...
                    > > > >
                    > > > > 1. Web Analytics Demystified laid out one big pro for me, which
                    > is
                    > > > > understanding where, geographically, our visitors come from
                    > (esp
                    > > > since
                    > > > > we're working on outreach to other countries). It's my
                    > understanding
                    > > > > that web server logs don't accurately count geography due to
                    > proxy
                    > > > > caching. Maybe it's true for page tags, but I didn't get that
                    > from
                    > > > the
                    > > > > book.
                    > > > >
                    > > > > 2. We're a google-friendly nonprofit (in other words, we've got
                    > a
                    > > > > google grant, so the more google we use, the better. :))
                    > > > >
                    > > > > 3. In our case, page tagging is a much easier setup than server
                    > > > logs,
                    > > > > and certainly cheaper than buying a web analytics product (FWIW
                    > at
                    > > > > former jobs I've used Webtrends - horrible processing
                    > experience -
                    > > > and
                    > > > > Urchin pre-Google - great product, but still processing large
                    > files
                    > > > > was a PITA).
                    > > > >
                    > > > > --- In webanalytics@yahoogroups.com, "Rachel" <remtheory@>
                    > wrote:
                    > > > > >
                    > > > > > Hey guys, I wanted a visual to illustrate why page tags
                    > (like
                    > > > google
                    > > > > > analytics) are better than web server logs because you have
                    > to go
                    > > > > > through the ISP to get visitor information instead of right
                    > from
                    > > > the
                    > > > > > browser (and therefore improving accuracy). First off, is
                    > what I
                    > > > drew
                    > > > > > even right? Secondly, is it easy to follow? I'm showing this
                    > > > during a
                    > > > > > brownbag about web analytics at my org. I added it as a file
                    > in
                    > > > the
                    > > > > > forum: http://tinyurl.com/2bx98b Thanks! Rachel
                    > > > > >
                    > > > >
                    > > >
                    > >
                    >
                  • Eric Peterson
                    Rachel, I humbly accept your apology, no harm done. As I said in my last response I can see how the chapter on log files may have created your confusion, and
                    Message 9 of 25 , Nov 1, 2007
                      Rachel,

                      I humbly accept your apology, no harm done. As I said in my last
                      response I can see how the chapter on log files may have created your
                      confusion, and I will work diligently to clarify that section with the
                      second edition of the book.

                      Thanks for the nice comments about the book!

                      Sincerely,

                      E.


                      --- In webanalytics@yahoogroups.com, "Rachel" <remtheory@...> wrote:
                      >
                      > Oh wow, I come back from 3 hours of meetings and I see an email from
                      > Eric saying I called him a liar and a bunch of new posts on this
                      > subject. This is exactly why email communication can be so difficult
                      > and frustrating.
                      >
                      > I in *no way* meant that I thought he was a liar. It was sarcasm. But
                      > how would he, or anyone on this list, have known that? I am truly
                      > sorry for the miscommunication. I love the book and have found it to
                      > be the only book on web metrics that I have not only managed to get
                      > through (numbers scare me, can you tell?), but have really enjoyed.
                      >
                      > I'll scurry away in my confusion over page tags and hope for the best.
                      >
                      > Be well,
                      > Rachel
                      >
                      > --- In webanalytics@yahoogroups.com, "Paul Holstein" <paul@> wrote:
                      > >
                      > > I wouldn't say he lied. I read a couple of his books and I know Eric
                      > > personally. I've never gotten the impression that he's said anything
                      > > other than what I've written.
                      > >
                      > > I'm sure he'll clarify this shortly.
                      > >
                      > > --Paul
                      > >
                      > >
                      > > --- In webanalytics@yahoogroups.com, "Rachel" <remtheory@> wrote:
                      > > >
                      > > > Really? So Eric Peterson lied? Or did I misread? :(
                      > > >
                      > > > --- In webanalytics@yahoogroups.com, "Paul Holstein" <paul@> wrote:
                      > > > >
                      > > > > Unfortunately, neither log files nor page tags gather
                      > > geographical
                      > > > > information. Both gather the IP address only. There is only one
                      > > IP
                      > > > > address sent to the web server and the client's Proxy servers
                      > > will
                      > > > > show the same one, no matter where the connection originated.
                      > > > >
                      > > > > To get geographical information, either solution must reference a
                      > > > > lookup table that tries to match up IP addresses with locations.
                      > > > > This is an imprecise science that is not 100% accurate.
                      > > > >
                      > > > > A page tagging solution may make it easier to gather geographic
                      > > > > information, but there is no reason to believe it would be more
                      > > > > accurate.
                      > > > >
                      > > > > --Paul
                      > > > >
                      > > > >
                      > > > >
                      > > > > --- In webanalytics@yahoogroups.com, "Rachel" <remtheory@> wrote:
                      > > > > >
                      > > > > > Wow, you guys are great. I knew it wasn't right, but I was
                      > > > > struggling
                      > > > > > to get the picture in my head. I think it's a bit clearer now.
                      > > For
                      > > > > > what it's worth, I'd like to explain my bias for page tagging.
                      > > Not
                      > > > > > trying to have a war, but my rationale is below. At this point
                      > > for
                      > > > > us
                      > > > > > it's not about choosing which method, it's just about giving
                      > > the
                      > > > > staff
                      > > > > > a brief understanding as to why we chose it. I'd like to make
                      > > the
                      > > > > > claim of greater accuracy and I thought I could do it based on
                      > > my
                      > > > > > reading of Web Analytics Demystified. So, here's the
                      > > rationale...
                      > > > > >
                      > > > > > 1. Web Analytics Demystified laid out one big pro for me, which
                      > > is
                      > > > > > understanding where, geographically, our visitors come from
                      > > (esp
                      > > > > since
                      > > > > > we're working on outreach to other countries). It's my
                      > > understanding
                      > > > > > that web server logs don't accurately count geography due to
                      > > proxy
                      > > > > > caching. Maybe it's true for page tags, but I didn't get that
                      > > from
                      > > > > the
                      > > > > > book.
                      > > > > >
                      > > > > > 2. We're a google-friendly nonprofit (in other words, we've got
                      > > a
                      > > > > > google grant, so the more google we use, the better. :))
                      > > > > >
                      > > > > > 3. In our case, page tagging is a much easier setup than server
                      > > > > logs,
                      > > > > > and certainly cheaper than buying a web analytics product (FWIW
                      > > at
                      > > > > > former jobs I've used Webtrends - horrible processing
                      > > experience -
                      > > > > and
                      > > > > > Urchin pre-Google - great product, but still processing large
                      > > files
                      > > > > > was a PITA).
                      > > > > >
                      > > > > > --- In webanalytics@yahoogroups.com, "Rachel" <remtheory@>
                      > > wrote:
                      > > > > > >
                      > > > > > > Hey guys, I wanted a visual to illustrate why page tags
                      > > (like
                      > > > > google
                      > > > > > > analytics) are better than web server logs because you have
                      > > to go
                      > > > > > > through the ISP to get visitor information instead of right
                      > > from
                      > > > > the
                      > > > > > > browser (and therefore improving accuracy). First off, is
                      > > what I
                      > > > > drew
                      > > > > > > even right? Secondly, is it easy to follow? I'm showing this
                      > > > > during a
                      > > > > > > brownbag about web analytics at my org. I added it as a file
                      > > in
                      > > > > the
                      > > > > > > forum: http://tinyurl.com/2bx98b Thanks! Rachel
                      > > > > > >
                      > > > > >
                      > > > >
                      > > >
                      > >
                      >
                    • Lothaire Ruellan
                      ... Speaking of obtaining geographical information from WA tools. Does anybody know why Webtrends ranks Virginia in the Top 3 of my traffic (presumably because
                      Message 10 of 25 , Nov 8, 2007
                        --- In webanalytics@yahoogroups.com, "chonchobar" <jmooney@...> wrote:
                        >
                        > Re: geography
                        > As someone noted, unless they're using telepathy - and elaborating on
                        > Paul's reply below - Google Analytics (or Webtrends, Omniture, etc.,
                        > etc.) is till using plain old HTTP log files to compile their reports

                        Speaking of obtaining geographical information from WA tools. Does
                        anybody know why Webtrends ranks Virginia in the Top 3 of my traffic
                        (presumably because of the AOL proxies located in that state), whereas
                        it is in my lower Top 10 in Google Analytics? Is google analytics
                        filtering AOL ip's from it's IP lookup and lumps them into the "not
                        set" category ?

                        Lothaire
                      • Paul Holstein
                        Most likely, Google has filtered them. It s easy to do. Wikipedia has a list of IPs from AOL. You can also get the list directly from AOL:
                        Message 11 of 25 , Nov 9, 2007
                          Most likely, Google has filtered them. It's easy to do. Wikipedia
                          has a list of IPs from AOL. You can also get the list directly from
                          AOL: http://webmaster.info.aol.com/proxyinfo.html

                          --Paul


                          --- In webanalytics@yahoogroups.com, "Lothaire Ruellan" <lruellan@...>
                          wrote:
                          >
                          > --- In webanalytics@yahoogroups.com, "chonchobar" <jmooney@> wrote:
                          > >
                          > > Re: geography
                          > > As someone noted, unless they're using telepathy - and elaborating on
                          > > Paul's reply below - Google Analytics (or Webtrends, Omniture, etc.,
                          > > etc.) is till using plain old HTTP log files to compile their reports
                          >
                          > Speaking of obtaining geographical information from WA tools. Does
                          > anybody know why Webtrends ranks Virginia in the Top 3 of my traffic
                          > (presumably because of the AOL proxies located in that state), whereas
                          > it is in my lower Top 10 in Google Analytics? Is google analytics
                          > filtering AOL ip's from it's IP lookup and lumps them into the "not
                          > set" category ?
                          >
                          > Lothaire
                          >
                        Your message has been successfully submitted and would be delivered to recipients shortly.