Miklos A' Vasarhelyi

1
Measuring the Web

• Miklos A. Vasarhelyi

2
A New Set of Measurements is evolving

• It has dynamic characteristics
• It is quite different from traditional measures
• The ethical implications are particularly
  daunting
• Laws are very different country to country
• Interesting cross-product implications

3
Who knows what?

• The merchant
• click path
• financial data
• personal data
• buying habits
• link to other databases?

• The user
• cookies
• tmp files
• files

• The ISP
• DNS
• logs
• click paths

• Net Nodes
• packets

4
Cookies

• What are cookies?
• Are pieces of information that foreign Web sites
  leave in the users computer for future
  identification, information, etc.
• The standards are contingent on the browser used
  (e.g. Netscape / Internet Explorer)
• Some security features apply (not too well)

5
Netscape HTTP Cookie File http//www.netscape.
com/newsref/std/cookie_spec.html This is a
generated file! Do not edit. www.etrade.com FA
LSE /cgi-bin/gx.cgi FALSE 931660324 etmember www.h
bo.com FALSE /schedule/bin FALSE 951267600 hbo .v
alueclick.com TRUE /cgi-bin/ FALSE 926286999 bc a0
010162 e12.zdnet.com8080 FALSE /clear FALSE 94668
4799 cgversion .zdnet.com TRUE / FALSE 946684799 b
rowser 0C4F013 .preferences.com TRUE / FALSE 11821
40421 PreferencesID www.businessweek.com FALSE /
FALSE 942189160 NGUserI e-care.att.com TRUE / FALS
E 946684799 CustId T.036793 .altavista.com TRUE /
FALSE 946641600 AV_UID d498199f .quicken.com TRUE
/ FALSE 1262390399 TID 234365523 .doubleclick.net
TRUE / FALSE 1920499140 id 76ea42fet4hf c
6
The User

• Identified or not identified to the Net
• the pentium III imbroglio
• Cookies
• PC media undercarriage a la Nielsen
• Many forms of identification (willing or
  unwilling)

7
The ISP

• May control the Domain Name Server
• if not, there is an additional player
• Has most information of any of the elements of
  the chain
• Strong ethical rules, and enforcement, must
  eventually arise

8
The Hop

• Many hops are the path of most Web activities
• Packets by themselves are of limited value
• Spoofing may be generated in any hop
• Busy hops drop packets
• Lost packets are foten retransmitted
• Certain protocols are tolerant

9
The Commerce Site

• separation of functions / secure / less secure
• identifiability of users
• IP identification
• ppp / slip
• make site password protected
• password management pros and cons
• click tracking
• usability
• linkage with databases / scanner data/ financial
  data

10
Conclusions

• For many reasons Web activity measurement is a
  major issue
• Great new opportunities for Marketeers
• Great issues with privacy
• Managers must have a wide view of the
  technological issues