>To: [log in to unmask] (James B. Morrison, Jr.)
>From: [log in to unmask] (James B. Morrison, Jr.)
>Subject: Internet Marketing Digest #0446
>Cc:=20
>Bcc:=20
>X-Attachments:=20
>
>>Return-path: <[log in to unmask]>
>>Date: Mon, 11 Sep 1995 10:46:10 -0700
>>From: Internet Marketing discussion list <[log in to unmask]>
>>Subject: Internet Marketing Digest #0446
>>Sender: [log in to unmask] (Glenn Fleishman)
>>X-Sender: [log in to unmask]
>>To: [log in to unmask]
>>Comments: Point of Presence Company, Seattle, Washington  <[log in to unmask]>
>>Comments: Point of Presence Company
>>
>>Internet Marketing Discussion List
>>
>>Digest #0446
>>
>>----------------------------------------------------------------------
>>The Internet Marketing Discussion List is sponsored by
>>Okidata                        <http://www.okidata.com>
>>and Downtown AOL               <http://downtown.web.aol.com>
>>----------------------------------------------------------------------
>>
>>To UNSUBSCRIBE send any text to the email address [log in to unmask]
>>
>>Send posts to [log in to unmask]
>>
>>This list is moderated by Glenn Fleishman <[log in to unmask]>
>>For info about the list, send INFO INTERNET-MARKETING to=
 [log in to unmask]
>>
>>Complete list archives: <http://www.popco.com/hyper/internet-marketing/>
>>For sponsorship info, add "sponsor.html" to that address
>>
>>----------------------------------------------------------------------
>>In this digest:
>>Re: Measuring Site Traffic ([log in to unmask] (Ed Hott))
>>Re: Client removed from listing due to pressure ([log in to unmask] (Peter
>>Arguelles))
>>Psychographics and the Web ("Steve Krause" <[log in to unmask]>)
>>GVU /HERMES Survey and Online Users (long (Sunil Gupta <[log in to unmask]>)
>>Re: Does Marketing own the Internet? (Nick Gassman=
 <[log in to unmask]>)
>>Re: Where to Place Your Ads? ([log in to unmask] (Debbie Reed))
>>Re: Measuring Site Traffic ([log in to unmask] (Brent Halliburton))
>>----------------------------------------------------------------------
>>
>>Date: 10 Sep 1995 08:22:23 -0700
>>From: [log in to unmask] (Ed Hott)
>>Subject: Re: Measuring Site Traffic
>>
>>
>>With all due respect to the ongoing discussion about state systems, there
>>are ways to mine significant information out of standard log files without
>>modifying the server or writing custom CGI scripts. Our product, Interse'
>>market focus, accepts log files in Common Logfile Format (CLF) and
>>approximates user sessions with *no* modifications to the server
>>environment. The full details are on our web site (www.interse.com).
>>Relevant to this discussion, here's how we define a "user".
>>
>>[ extracted from www.interse.com/marketfocus/user.html ]
>>
>>We define a user as a single session or interaction with a web site. In
>>other words, if Neal visits our web site on Monday, and then returns on
>>Tuesday, he is counted as two users. The privacy of the individual users=
 is
>>ensured, as you won't be able to distinguish Neal from Debra who also
>>visited your site on Monday.
>>
>>To differentiate user sessions, our user algorithm analyzes the request
>>stream in three passes:
>>
>>1. Separates requests into groups of similar "differentiating
>>characteristics." Requests from a single user session always use the same
>>browser, originate from the same Internet address, use the same web=
 protocol
>>and request format, etc.
>>
>>2. Sorts requests within each group chronologically. If there is a=
 25-minute
>>time gap in the request stream, then the previous user session is=
 considered
>>complete.
>>
>>3. If a page is requested and not cached, and there is a high probability
>>that this page is in the cache of the current session, then the current
>>request is considered part of a new user session.
>>
>>Our user algorithm is conservative, underestimating a sites' user session
>>count, but never overestimating it. The algorithm is consistent, so
>>comparisons between time periods and even between sites are valid. The
>>results enable you to make more informed Internet business decisions.
>>
>>[ end of extract ]
>>
>>The orignal post was seeking "more relevant and in-depth information than
>>the "hit-list" generated by the ISP". Interse' market focus is an
>>off-the-shelf solution that lets you analyze site usage using standard log
>>files. There are server side approaches as well, but if you're using an=
 ISP
>>for web hosting services, don't forget that you may have to convince your
>>ISP to modify *his/her* server environment to accommodate *your*=
 customers.
>>
>>Cheers,
>>
>>Ed
>>_________________________________________________
>>
>>Ed Hott
>>Interse' Corporation
>>[log in to unmask]
>>408 732-0932  ext. 233
>>408 732-7038  fax
>>http://www.interse.com
>>
>>----------------------------------------------------------------------
>>
>>Date: 10 Sep 1995 08:23:53 -0700
>>From: [log in to unmask] (Peter Arguelles)
>>Subject: Re: Client removed from listing due to pressure
>>
>>
>>Sharon Lane ([log in to unmask]) wrote:
>>
>>>Today the directory service called my client and told him that they had
>>>removed his ad. They had received complaints from competitors, that did=
 not
>>>want his type of service offered on a directory where they listed their
>>services.
>>
>>Hullo all!
>>
>>I have recently been wooed by several ISP's to move my website to their
>>service. A couple of days ago I received a note from one presenting me=
 with
>>a URL for a "Site of ill repute" (tm). The site
>><http://www.servint.com/online> is located directly off my ISP's base URL=
 as
>>am I. The not-so-concealed implication was that both the ISP and my=
 company
>>would be guilty by association for sharing the same base URL with this=
 other
>>company.
>>
>>Now, I have nothing against the site in question, nor any site of its ilk,
>>but I do believe that others might -- specifically, my potential=
 customers.
>>So I wrote to my ISP's administration and suggested that those other guys
>>should be offered a domain of their own, free of charge if necessary. I=
 have
>>previously applied for my domain <http://www.xebec.com> but Internic seems
>>to have lost it...
>>
>>I think this relates to Sharon's question, and my reactions surely=
 elaborate
>>my own opinion.
>>
>>Any others?
>>
>>~Pete
>>
>>* Laser Toners * Ribbons * Inkjets * http://www.servint.com/Xebec
>>=3D \=3D=3D /=3D=3D=3D ____/=3D __ \=3D=3D=3D ___/=3D ___/=3D=3D=3D=3D=3D =
  1-800-429-0123   =3D=3D=3D=3D=3D
>> =3D \ /=3D=3D=3D /=3D=3D=3D=3D=3D __/ /=3D=3D /=3D=3D=3D=3D /=3D =
 310-827-0123 Fax 310-305-8185
>>  =3D  =3D=3D=3D ___/=3D=3D ___ <=3D=3D __/=3D=3D /=3D=3D X E B E C  C O R=
 P O R A T I O N
>> =3D / \=3D /=3D=3D=3D=3D=3D ___/ /=3D /=3D=3D=3D=3D  \=3D=3D=3DOnline=
 Imaging/Supply Superstore
>>=3D_/=3D=3D_\____/=3D______/=3D____/=3D=3D\____/=3D   email=
 [log in to unmask]
>>
>>----------------------------------------------------------------------
>>
>>Date: 10 Sep 1995 08:25:08 -0700
>>From: "Steve Krause" <[log in to unmask]>
>>Subject: Psychographics and the Web
>>
>>
>>In Internet Marketing Digest #0444, Glenn Fleishman
>><[log in to unmask]> wrote:
>>
>>> I just had an interesting conversation with a marketing person
>>> at a design firm who had just compelted a refresher course at
>>> the local university. She that traditional demographic categories
>>> have collapsed, since you can't make assumptions about people
>>> based on their age, gender, etc., any more that are reliable. So
>>> instead, marketers use fuzzy conceptual groups and "psychographics,"
>>> where you group people by what they think about things, rather than
>>> their physical, racial, or other characteristics.
>>
>>Psychographics are indeed useful tools, although they are often
>>most useful when augmenting--not replacing--demographics. The relative
>>value of the two in predicting consumer behavior tends to depend on
>>the particular behavior (and related products/services) you're trying
>>to measure.
>>
>>My group at SRI International believes that once site providers get past
>>the current infatuation with counting hits, they'll increasingly need
>>psychographics to actually make hits count--that is, to insert the most
>>appropriate and effective ads, to offer links with the best chance of
>>being clicked, and to customize a site's presentation for individuals'
>>affinities.
>>
>>Our site is at <http://future.sri.com> for those interested. It includes
>>information about iVALS and VALS 2. Initially, we applied the latter
>>(a general-purpose psychographic segmentation for the U.S. population)
>>to the Web audience, surveying 5500 unique respondents earlier this year.
>>Some of the lessons we learned from that exercise are available in a
>>paper on the site. One meta-lesson was that we needed an Internet-
>>specific segmentation, and thus begat iVALS.
>>
>>- --Steve
>>
>>P.S. The site allows you to type yourself for both iVALS and VALS 2.
>>     (VALS is an acronym for Values and Lifestyles, which is a program
>>     that has been creating commercial psychographic systems since the
>>     late 1970s. VALS is a trademark of SRI International.)
>>
>>----------------------------------------------------------------------
>>
>>Date: 10 Sep 1995 08:26:01 -0700
>>From: Sunil Gupta <[log in to unmask]>
>>Subject: GVU /HERMES Survey and Online Users (long)
>>
>>
>>In recent postings Elizabeth Lane Lawley has expressed reservations
>>regarding the GVU/HERMES study=EDs usefulness in describing the "online
>>community as a whole".  More specifically she stated, "power internet
>>users, maybe, but the "average consumer" who wanders onto the web from AOL
>>or their local internet provider? No."
>>
>>As a matter of fact, the survey does include considerable information=
 about
>>such users (who also surf the Web).  Of the 13,006 WWW users responding to
>>the 3rd survey, 6,800 subscribed to either AOL, CompuServe, or Prodigy.
>>Listed below, are some of the differences between these 6,800 and the
>>others. (please keep in mind that the 3rd survey was conducted on the Web
>>between April 10 - May10, and not all of the services had web browsers for
>>the entire duration of the survey.  Thus, whatever conclusions you wish to
>>reach can apply only to the Web users of these services as of the survey
>>date.  Please also keep all of the other caveats regarding the study in
>>mind).
>>
>>                          Non-subscribers       Subscribers
>>DEMOGRAPHIC
>>Income < $50,000                55%             35%
>>Single                          50%             31%
>>Education - Ph.D.=EDs             10%             5%
>>Occupation -    Managerial      7%              17%
>>                Professionals   17%             26%
>>                Educ related    34%             14%
>>
>>COMPUTER RELATED
>>Internet Access thru Edu        44%             10%
>>Modem Speed     >1m             32%             13%
>>                14k             33%             56%
>>Platform -      Windows         35%             67%
>>                Unix            16%             2%
>>Less than 1 year on Internet    35%             64%
>>Surf once a day or more         80%             64%
>>>50 item in hotlist             36%             29%
>>Find Web sites thru
>>        Magazines               59%             71%
>>        Newsgroups              64%             52%
>>        .sigs                   38%             28%
>>Primary use of Web
>>        Academic Research       42%             26%
>>        Business Research       34%             43%
>>        Educational             50%             43%
>>Not willing to pay fees
>>for Web resources               26%             20%
>>
>>COMMERCIAL STUFF
>>Use commercial WWW sites for
>>purchase information            53%             63%
>>Bought through any online source
>>        Apparel                 2%              7%
>>        Books                  14%              15%
>>        Casual Clothes          2%              6%
>>        Electronics < $50       3%              6%
>>        Electronics > $50       3%              6%
>>        Gourmet Food            2%              3%
>>        Hardware < $50          9%              19%
>>        Hardware > $50          10%             17%
>>        Stocks/Bonds            5%              12%
>>        Music                   11%             11%
>>        Software < $50          21%             35%
>>        Software > $50          13%             23%
>>        Travel related          6%              15%
>>
>>Sought information through any online source
>>        Apparel                 9%              13%
>>        Books                   43%             42%
>>        Casual Clothes          6%              10%
>>        Concerts/Plays          18%             18%
>>        Electronics < $50       20%             27%
>>        Electronics > $50       27%             37%
>>        Gourmet Food            10%             12%
>>        Hardware < $50          48%             58%
>>        Hardware > $50          62%             72%
>>        Stocks/Bonds            23%             35%
>>        Movies/Videos           34%             36%
>>        Music                   36%             39%
>>        Software < $50          59%             65%
>>        Software > $50          63%             70%
>>        Travel related          35%             49%
>>
>>Quick summary: The online service users who surf the Web are richer,
>>non-academics, newer to the internet, and are more likely to be interested
>>in using the Web for purchase and shopping related information across many
>>product categories.
>>
>>Details for individual online services and other survey
>>questions are available by downloading the original datasets
>>(http://www.cc.gatech.edu/gvu/user_surveys/survey-04-1995/datasets/),
>>and then running the analysis using your own analysis program. A simpler
>>procedure for customized analysis will be available on September 27th.
>>Comments and suggestions for the 4th survey are welcome.
>>
>>Sunil Gupta
>>Director, HERMES Project
>>University of Michigan Business School
>>http://www.umich.edu/~sgupta/hermes/
>>[log in to unmask]
>>
>>----------------------------------------------------------------------
>>
>>Date: 10 Sep 1995 08:26:47 -0700
>>From: Nick Gassman <[log in to unmask]>
>>Subject: Re: Does Marketing own the Internet?
>>
>>
>>Larry Thomson said
>>
>>*Quote
>>Debate seems to be raging among my peers as to what functional=
 organization
>>within a company is most responsible for bringing the Internet to their
>>organizations and who owns it after it's implemented.
>>
>>*snip
>>
>>Would also like to know where the list members think responsibility for
>>Customer Support belongs within a business organization.  For this post I
>>would define customer support as all communication to and from the=
 customer
>>including product information, press releases, software patches,
>>configuration and support information, troubleshooting info etc
>>
>>*end quote
>>
>>There seems to be some background to this question that we are not
>>party to. I imagine Larry has in mind an organisation of a particular
>>size, and from the last paragraph he seems to be talking about a
>>software company??? As a general point to list contributors, can I
>>suggests that such assumptions should be made clear, as the answer in
>>this case can only be 'it depends'.
>>
>>What it depends on are such factors as the size of the organisation,
>>the way it is structured, and the use they wish to make of the
>>Internet.
>>
>>If we assume a large enough organisation to have the full range of
>>functional departments, then I believe that ownership should not lie
>>with any one department, but with a corporate area, if such an area
>>exists. In a large organisation there will be many business units
>>with an interest in the use of the Internet, and there is a danger if
>>one department owns it that not all interests will be adeqautely
>>represented.
>>
>>My view is that 'ownership' should lie with a central dept and for
>>there to be a steering board of representatives from the business
>>areas affected.
>>
>>As to who manages contact, the answer should be that all the relevant
>>areas do. It doesn't have to be one area only. Have a look at
>>http://www.tesco.co.uk
>>They have a contact section which lists email addresses for customer
>>relations, investor relations, computers for schools, webmaster. I
>>think this is the way to go, and I haven't yet seen a lot of it Any
>>department with legitimate reasons for customer contact should talk
>>direct to the customer.
>>
>>Multimedia across the Internet is here
>>http://www.route-one.co.uk/route-one
>>Nick Gassman known as [log in to unmask]
>>
>>----------------------------------------------------------------------
>>
>>Date: 10 Sep 1995 11:06:50 -0700
>>From: [log in to unmask] (Debbie Reed)
>>Subject: Re: Where to Place Your Ads?
>>
>>
>>Although in the formative stages, Nielsen, SRDS, and other traditional
>>audience research firms are looking at the problem (opportunity) you
>>discuss.  In the meantime, it seems the only track is a "do it yourself" -
>>exactly what you're doing - getting data from individual sites.
>>
>>The Chamber of Commerce would be a good start.  Also city listings, small
>>business sites, home based business publications.   And don't forget the
>>opportunity of posting to appropriate newsgroups in your area - with
>>disclaimers if you are selling your services.  Don't overlook traditional
>>media forms to build name awareness:  local newspaper, business journals,
>>radio.  Speaking opportunities as part of an overall integrated program=
 will
>>also produce results.
>>
>>What you are suggesting re: online help is an excellent idea;  in a sense,
>>the formation of rep firms handling areas, site types, i. e., electronic
>>journals (consumer & trade), geographic areas, specialties such as=
 science,
>>education, computers, marketing and leisure activities such as recreation,
>>travel, etc.  Master list development could begin in similar fashion to=
 the
>>volumes of SRDS or take a geographic approach like Gale's directories.
>>Online databases for searching, or printed reports by subscription.
>>
>>We've been looking at this same situation locally (New Orleans) and have
>>categorized type information as mentioned above.  Sites are approached
>>individually; there are no standard costs.  Everything appears=
 experimental
>>at the moment based on hits; psychographic information is assumed by=
 content.
>>
>>Debbie
>>
>>****************************************************************
>>DRI Media                       --      504-365-1401
>>405 Gretna Blvd., Suite 212     --      504-365-1433 Fax
>>Gretna, LA  70053               --      [log in to unmask]
>>- ----Advertising Research, Media Buying, Internet Marketing----
>>****************************************************************
>>
>>----------------------------------------------------------------------
>>
>>Date: 10 Sep 1995 13:25:20 -0700
>>From: [log in to unmask] (Brent Halliburton)
>>Subject: Re: Measuring Site Traffic
>>
>>
>>Ed wrote about Interse's Market Focus:
>>
>>>To differentiate user sessions, our user algorithm analyzes the request
>>>stream in three passes:
>> [ - snip - ]
>>>2. Sorts requests within each group chronologically. If there is a=
 25-minute
>>>time gap in the request stream, then the previous user session is=
 considered
>>>complete.
>> [ - snip - ]
>>>Our user algorithm is conservative, underestimating a sites' user session
>>>count, but never overestimating it. The algorithm is consistent, so
>>>comparisons between time periods and even between sites are valid. The
>>>results enable you to make more informed Internet business decisions.
>>
>>Ed is right, due to limitations of the http protocol, without modifying=
 the
>>server environment it is impossible to do little more than estimate the
>>amount of users.  If a company maintaining a web site is unable to modify
>>their server environment, than they are forced to rely on a series of
>>algorithms to judge what pages a user looks at.
>>
>>I have a lot of questions about this "rule of thumb" approach.  I assume
>>from your discussion that the way you are "conservative" and
>>"underestimate" a user session count is by having a 25 minute time out
>>period instead of something like 15 minutes.  However, that seems
>>inaccurate.  I am relatively certain that every hit to the server gets
>>allocated somewhere.  You can either allocate it to distinct users, or in
>>your conservative approach, assign them to the same user.  What is a worse
>>crime?  To have 2 users where there was one, or to have one user making
>>twice as many hits as he would have.  To call this conservative would be
>>one way to describe it.  I think it was simply a design decision when
>>attempting to determine how to err when one recognizes the possibility of
>>inaccuracy.  To illustrate this, I flipped open my (Common Log Format) log
>>files.  Here we have part of a session from a visitor from Netscape.  He
>>looked at about 20 pages before coming to this.
>>
>>unknown.netscape.com - - [01/Sep/1995:16:01:49 -0400] "GET /iis/ HTTP/1.0"
>>200 809
>>unknown.netscape.com - - [01/Sep/1995:16:02:04 -0400] "GET /groupcortex/
>>HTTP/1.0" 200 1639
>>unknown.netscape.com - - [01/Sep/1995:16:02:18 -0400] "GET
>>/groupcortex/news/news.html HTTP/1.0" 200 4425
>>unknown.netscape.com - - [01/Sep/1995:16:29:02 -0400] "GET /contact.html
>>HTTP/1.0" 200 1765
>>
>>This illustrates the question of algorithms.  He was gone more than 25
>>minutes so he would be counted as two sessions.  Common log format log=
 file
>>analysis would indicate that it was two distinct users requesting on
>>average 10 pages a session, rather than indicating that it was one user=
 who
>>clicked on the link to read about our Interactive Age article and then=
 came
>>back to fill out the contact page, which is what actually took place.  One
>>user, 20 pages, and other information.
>>
>>The common log format does not even contain much of the information that=
 is
>>important to improve the efficacy of the algorithm.  For example, there is
>>no user-agent field in common log formats.
>>
>>That is why state environments are so important.  They are the only way to
>>eliminate reliance on estimates to count people.  They give you a hard
>>count.
>>
>>Even state environments can be built that are more or less effective.  The
>>manner in which you track users can have significant bearing on how much
>>information is extracted from the tracking process.
>>
>>How do other people feel about this question?  I know that up until we
>>built a state environment and saw how effective it was we relied on
>>algorithms.  Now I think that tools like Interse's are an attempt to
>>provide a UI to information, but I don't expect them to be accurate=
 because
>>they simply don't have access to the information and data necessary to
>>provide accuracy.
>>
>>
>>Brent Halliburton
>>
>>President                                       http://www.cortex.net/
>>Group Cortex                               2300 Chestnut St. Suite 230
>>                                                Philadelphia, PA 19103
>>[log in to unmask]                 Phone 215.854.0646   Fax 215.854.0665
>>----------------------------------------------------------------------
>>
>>----------------------------------------------------------------------
>>Our Sponsors:
>>
>>Okidata <http://www.okidata.com>
>>Okidata is a 23 year old company that's built its success on meeting
>>the computer printing needs of small to medium sized organizations
>>throughout North and South America
>>
>>Downtown AOL <http://downtown.web.aol.com>
>>The streets of Downtown AOL are populated with an eclectic group of
>>merchants, offering products and services that span the spectrum from
>>novelty mousepads to regional gourmet food items.
>>----------------------------------------------------------------------
>>
>>
>>
>>
>