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

Re: [webanalytics] site logs vs. javascript

Expand Messages
  • Steve
    Couple of thoughts: * Have you tagged *all* your pages? Can possibly use some sort of simple site mirror tool and then find all those pages that don t have the
    Message 1 of 8 , Jul 11, 2006
    • 0 Attachment
      Couple of thoughts:
      * Have you tagged *all* your pages? Can possibly use some sort of
      simple site mirror tool and then find all those pages that don't have
      the tagging code.

      * What sort of demographic are you hitting? Are they using browsers
      that are likely to have javascript disabled or not implemented (this
      one hurts me personally on my own site - get a lot of folk using very
      simple browsers: wget, lynx etc)

      * There may be some issues with how the hosting company had configured
      WT. It's been a while for myself, but I've managed to cause WT to
      *vastly* inflate numbers pretty easily. Like order of magnitude
      difference. The assumptions it makes and I make being quite different.

      * I'm current in the process of working with another group with their
      domino logs and such. I haven't found anything too glaringly different
      from "normal" log analysis. Hmm actually there is one now I think some
      more - Those domino servers run a *lot* of virtual hosts. For reasons
      unknown to us, all the virtual hosts are lumped in the one log. I've
      written a "not even as hard as trivial" perl script to split those out
      for processing. Perhaps your providor hadn't configured WT to handle
      the virtualisation, so you've been counting other sites???

      * Check your filters? Do you have any filters in GA that aren't in WT?

      * Just doing a quick compare with my own site's logs vs GA - GA is
      about 50% of what the pure log analysis reports. I have looked in
      detail in the past and GA is more correct. But GA is under for the
      afore mentioned reason with browsers. The real number is closer to a
      third up from what GA reports.

      * In my personal observation, as a site gets busier, traffic from
      spiders and such tends to not hurt visitor numbers so much - gradually
      becomming a smaller percentage. But for small, not so busy sites,
      spiders can totally swamp visit(or) numbers. YMMV.


      My personal feel is that your new numbers do sound about right. For a
      given value of "right". :-)


      The flip side to all the above: Does it *matter*!
      Are you suddenly experiencing a drop in revenue from the site? If not
      - well no it doesn't matter. If yes then maybe you've done something
      else to turn people off?


      Cheers!
      Steve


      On 7/12/06, Rachel <remtheory@...> wrote:
      > I just joined an org that is hosting their site using (gasp) lotus,
      > and the hosting company runs monthly webtrends reports (7.0, looks
      > like). Well, we recently redesigned the site are now using Google
      > Analytics and the numbers aren't just a little under, they are a LOT
      > under what webtrends is saying. We're talking half the number of
      > visits. I'm sure some of this is due to javascript being turned off
      > and people leaving pages before the whole thing loads, but I can't
      > imagine that that would apply to half the visitors. I wonder, too, how
      > many "visits" were counted due to people going to 404s, that wouldn't
      > be counted with Google. Are there any other factors, do you think,
      > that would contribute to such a drastic difference? Heck, I'm not even
      > sure if lotus is a factor, but I'd love to know if anyone's had
      > experience on that system.
    • Rachel
      Wow, thank you all SOO much. Steve, what a great overview, and Lars and Fred, thank you for your input. Right now I ve got no access to our hosting company,
      Message 2 of 8 , Jul 11, 2006
      • 0 Attachment
        Wow, thank you all SOO much. Steve, what a great overview, and Lars
        and Fred, thank you for your input. Right now I've got no access to
        our hosting company, but I'll make sure to ask these questions.
        Ultimately I'm hoping we go away from lotus to a true web cms... I
        hope I can use this as further justification to do so (i've already
        got a laundry list of reasons!).

        And, well, as far as less revenue, we're a nonprofit, so it's more
        about board confidence. So, I'm sure we'll still be using the numbers,
        but now we have to expose them for what they are... very flexible. ;)

        Rachel

        --- In webanalytics@yahoogroups.com, Steve <nuilvows@...> wrote:
        >
        > Couple of thoughts:
        > * Have you tagged *all* your pages? Can possibly use some sort of
        > simple site mirror tool and then find all those pages that don't have
        > the tagging code.
        >
        > * What sort of demographic are you hitting? Are they using browsers
        > that are likely to have javascript disabled or not implemented (this
        > one hurts me personally on my own site - get a lot of folk using very
        > simple browsers: wget, lynx etc)
        >
        > * There may be some issues with how the hosting company had configured
        > WT. It's been a while for myself, but I've managed to cause WT to
        > *vastly* inflate numbers pretty easily. Like order of magnitude
        > difference. The assumptions it makes and I make being quite different.
        >
        > * I'm current in the process of working with another group with their
        > domino logs and such. I haven't found anything too glaringly different
        > from "normal" log analysis. Hmm actually there is one now I think some
        > more - Those domino servers run a *lot* of virtual hosts. For reasons
        > unknown to us, all the virtual hosts are lumped in the one log. I've
        > written a "not even as hard as trivial" perl script to split those out
        > for processing. Perhaps your providor hadn't configured WT to handle
        > the virtualisation, so you've been counting other sites???
        >
        > * Check your filters? Do you have any filters in GA that aren't in WT?
        >
        > * Just doing a quick compare with my own site's logs vs GA - GA is
        > about 50% of what the pure log analysis reports. I have looked in
        > detail in the past and GA is more correct. But GA is under for the
        > afore mentioned reason with browsers. The real number is closer to a
        > third up from what GA reports.
        >
        > * In my personal observation, as a site gets busier, traffic from
        > spiders and such tends to not hurt visitor numbers so much - gradually
        > becomming a smaller percentage. But for small, not so busy sites,
        > spiders can totally swamp visit(or) numbers. YMMV.
        >
        >
        > My personal feel is that your new numbers do sound about right. For a
        > given value of "right". :-)
        >
        >
        > The flip side to all the above: Does it *matter*!
        > Are you suddenly experiencing a drop in revenue from the site? If not
        > - well no it doesn't matter. If yes then maybe you've done something
        > else to turn people off?
        >
        >
        > Cheers!
        > Steve
        >
        >
        > On 7/12/06, Rachel <remtheory@...> wrote:
        > > I just joined an org that is hosting their site using (gasp) lotus,
        > > and the hosting company runs monthly webtrends reports (7.0, looks
        > > like). Well, we recently redesigned the site are now using Google
        > > Analytics and the numbers aren't just a little under, they are a LOT
        > > under what webtrends is saying. We're talking half the number of
        > > visits. I'm sure some of this is due to javascript being turned off
        > > and people leaving pages before the whole thing loads, but I can't
        > > imagine that that would apply to half the visitors. I wonder,
        too, how
        > > many "visits" were counted due to people going to 404s, that wouldn't
        > > be counted with Google. Are there any other factors, do you think,
        > > that would contribute to such a drastic difference? Heck, I'm not
        even
        > > sure if lotus is a factor, but I'd love to know if anyone's had
        > > experience on that system.
        >
      • der_statman
        Hi Rachel, Yours is a very common frustration for those changing reporting systems. The first place I would check is your WebTrends configuration. It sounds
        Message 3 of 8 , Jul 11, 2006
        • 0 Attachment
          Hi Rachel,

          Yours is a very common frustration for those changing reporting
          systems.

          The first place I would check is your WebTrends configuration. It
          sounds like you were using the log file analysis product (as opposed
          to their tagged solution). A couple specifics:

          Check your filters to see if WebTrends was cutting out spider/robot
          traffic. If not, that could easily account for the lion's share of
          your difference, especially if you had some kind of site monitoring
          tool pinging the site. Spiders/robots/monitors will not load images
          or run javascript, thus they create a log file entry, but don't get
          counted by a javascript tag.

          Also, be sure that both systems (or neither, I suppose) are filtering
          out the same set of IP addresses, to make sure that you're comparing
          apples to apples.

          Basically, you're going through the configuration to see if there are
          any specific and significant types of traffic that shows up in
          WebTrends but not Google

          One thing to keep in mind, don't expect to ever be able to reconcile
          the two very different systems perfectly. There are inherent
          differences between the log file approach of WT and the tagged
          approach of GA. I'd say that if you can get the two to match within
          +/- 10%, you're doing pretty well. You might even be able to live
          with up to a 20% difference.

          HTH,
          AaronF

          --- In webanalytics@yahoogroups.com, "Rachel" <remtheory@...> wrote:
          >
          > Well, we recently redesigned the site are now using Google
          > Analytics and the numbers aren't just a little under, they are a LOT
          > under what webtrends is saying. We're talking half the number of
          > visits. I'm sure some of this is due to javascript being turned off
          > and people leaving pages before the whole thing loads, but I can't
          > imagine that that would apply to half the visitors. I wonder, too,
          how
          > many "visits" were counted due to people going to 404s, that wouldn't
          > be counted with Google. Are there any other factors, do you think,
          > that would contribute to such a drastic difference? Heck, I'm not
          even
          > sure if lotus is a factor, but I'd love to know if anyone's had
          > experience on that system.
        • Steve
          No worries Rachel. Glad to help. Being a non-profit you re more like the sites I look after - $ in and of itself isn t the primary target. I m sure I ll be
          Message 4 of 8 , Jul 12, 2006
          • 0 Attachment
            No worries Rachel. Glad to help.

            Being a non-profit you're more like the sites I look after - $ in and
            of itself isn't the primary target.

            I'm sure I'll be corrected by the more experienced members if I get
            this wrong.. :-)

            The trick now is to focus on what IS important. If you can accept that
            the numbers are being misinterpreted, you're well and truely ready for
            the next step: Making the numbers even smaller!

            Normally with the info style sites we focus on people - visitors. Hey
            we had X people hit the site today. Sounds good, easy to grasp. And
            pretty useless. :-)
            Explaining why it's useless is hard tho. Particularly if your explain
            must be condensed to a one line sentence with a single number in it.

            So you need to emphasise the actual real positive outcomes of the
            website. Most of the NGO's I've worked with in the past have been in
            the medical/health space. So to use some contrived examples:
            * Hey we helped X people learn more about AIDS today!
            Or:
            * We had an extra 25 people sign into our forums and start actively
            using them this week! But that's down a bit from the previous week -
            probably due to Uni holidays. Most of them used the general
            lung-cancer discussion forum.


            That's the sort of thing I suspect you'll find that an NGO board is
            really interested in. That the seriously limited resources they do
            have are being well used. By reporting pure people numbers there is a
            tendency to assume that those people actually used the site as it was
            intended. Sadly this is often not the case. :-)


            Welcome to the joy of Analytics.

            Cheers!

            Steve


            On 7/12/06, Rachel <remtheory@...> wrote:
            > Wow, thank you all SOO much. Steve, what a great overview, and Lars
            > and Fred, thank you for your input. Right now I've got no access to
            > our hosting company, but I'll make sure to ask these questions.
            > Ultimately I'm hoping we go away from lotus to a true web cms... I
            > hope I can use this as further justification to do so (i've already
            > got a laundry list of reasons!).
            >
            > And, well, as far as less revenue, we're a nonprofit, so it's more
            > about board confidence. So, I'm sure we'll still be using the numbers,
            > but now we have to expose them for what they are... very flexible. ;)
          • Arun Sharma
            hi, A fairly basic query on webanalytics on a group which seems to be in fairly advanced stages of web analytics knowledge. I am completely new to web
            Message 5 of 8 , Jul 12, 2006
            • 0 Attachment
              hi,
              A fairly basic query on webanalytics on a group which seems to be in fairly advanced stages of web analytics knowledge.
              I am completely new to web analytics and have to make a presentation to my team(equally new to web business) on how web analytics work. the primary objective is to understand the working rather than the appliactions for marketing. If someone could guide me on resources where I can find info on various technologies(like i read abt java scrpit vs log analysis) and basics of web analytics working and metrics, i shall be really thankful.
              looking forward to some responses
              Thanks
              Arun


              Send instant messages to your online friends http://uk.messenger.yahoo.com

              [Non-text portions of this message have been removed]
            • Rachel
              Thanks Steve, this is great advice. This discrepency has been the catalyst for upper management to revisit the way we report on web activity anyway, so it s
              Message 6 of 8 , Jul 12, 2006
              • 0 Attachment
                Thanks Steve, this is great advice. This discrepency has been the
                catalyst for upper management to "revisit the way we report on web
                activity" anyway, so it's a great time for us to do something like you
                suggest!

                Rachel

                --- In webanalytics@yahoogroups.com, Steve <nuilvows@...> wrote:
                >
                > No worries Rachel. Glad to help.
                >
                > Being a non-profit you're more like the sites I look after - $ in and
                > of itself isn't the primary target.
                >
                > I'm sure I'll be corrected by the more experienced members if I get
                > this wrong.. :-)
                >
                > The trick now is to focus on what IS important. If you can accept that
                > the numbers are being misinterpreted, you're well and truely ready for
                > the next step: Making the numbers even smaller!
                >
                > Normally with the info style sites we focus on people - visitors. Hey
                > we had X people hit the site today. Sounds good, easy to grasp. And
                > pretty useless. :-)
                > Explaining why it's useless is hard tho. Particularly if your explain
                > must be condensed to a one line sentence with a single number in it.
                >
                > So you need to emphasise the actual real positive outcomes of the
                > website. Most of the NGO's I've worked with in the past have been in
                > the medical/health space. So to use some contrived examples:
                > * Hey we helped X people learn more about AIDS today!
                > Or:
                > * We had an extra 25 people sign into our forums and start actively
                > using them this week! But that's down a bit from the previous week -
                > probably due to Uni holidays. Most of them used the general
                > lung-cancer discussion forum.
                >
                >
                > That's the sort of thing I suspect you'll find that an NGO board is
                > really interested in. That the seriously limited resources they do
                > have are being well used. By reporting pure people numbers there is a
                > tendency to assume that those people actually used the site as it was
                > intended. Sadly this is often not the case. :-)
                >
                >
                > Welcome to the joy of Analytics.
                >
                > Cheers!
                >
                > Steve
                >
                >
                > On 7/12/06, Rachel <remtheory@...> wrote:
                > > Wow, thank you all SOO much. Steve, what a great overview, and Lars
                > > and Fred, thank you for your input. Right now I've got no access to
                > > our hosting company, but I'll make sure to ask these questions.
                > > Ultimately I'm hoping we go away from lotus to a true web cms... I
                > > hope I can use this as further justification to do so (i've already
                > > got a laundry list of reasons!).
                > >
                > > And, well, as far as less revenue, we're a nonprofit, so it's more
                > > about board confidence. So, I'm sure we'll still be using the
                numbers,
                > > but now we have to expose them for what they are... very flexible. ;)
                >
              • Anil Batra
                Start by reading Web Analytics Demystified by Eric Peterson ( http://www.webanalyticsdemystified.com/about_wad.asp) and the archive of this news group. Also,
                Message 7 of 8 , Jul 12, 2006
                • 0 Attachment
                  Start by reading "Web Analytics Demystified" by Eric Peterson (
                  http://www.webanalyticsdemystified.com/about_wad.asp) and the archive of
                  this news group.
                  Also, keep following this newsgroup, you will learn a lot.

                  - Anil Batra


                  On 7/12/06, Arun Sharma <arunsharma_47@...> wrote:
                  >
                  > hi,
                  > A fairly basic query on webanalytics on a group which seems to be in
                  > fairly advanced stages of web analytics knowledge.
                  > I am completely new to web analytics and have to make a presentation to my
                  > team(equally new to web business) on how web analytics work. the primary
                  > objective is to understand the working rather than the appliactions for
                  > marketing. If someone could guide me on resources where I can find info on
                  > various technologies(like i read abt java scrpit vs log analysis) and basics
                  > of web analytics working and metrics, i shall be really thankful.
                  > looking forward to some responses
                  > Thanks
                  > Arun
                  >
                  >
                  > Send instant messages to your online friends http://uk.messenger.yahoo.com
                  >
                  > [Non-text portions of this message have been removed]
                  >
                  >
                  >



                  --
                  Anil Batra


                  [Non-text portions of this message have been removed]
                Your message has been successfully submitted and would be delivered to recipients shortly.