Federal Trade Commission
600 Pennsylvania Avenue
Washington, DC 20580
October 18, 1999
Comments in Regards to:
Online Profiling Project Docket No. 990811219-9219-01
This is a comment regarding the questions:
# 9 - "What are the cost and benefits, to the consumer and the industry, of online profiling."
# 6 - "Are there technologies in development that will enable the creation of consumer profiles that identify the individual consumer?"
Regarding Question # 9 - "What are the cost and benefits, to the consumer and the industry, of online profiling."
The industry and business processes are new and the "clickstream profiling" solution is likely many changes will occur before the "best" method develops.
It must be recognized that online commerce is in a very early stage of development and that the business processes and models are still incomplete. The industry has not proven that any particular method "works" and it is likely we will pass through several different consumer profiling models before a standard develops. As such any attempt at regulation now will likely be wasted effort, and perhaps counter productive because it may focus government's efforts on solving yesterday's problems.
Business Week, October 25, 1999, makes this comment (p. 146) about online profiler Engage and its anonymous user 35 million profiles, " it is too early to tell is Engage really works, "
Clickstream profiling does not work because it relies on incomplete information.
Clickstreams simply do not provide the vendor enough "actionable" information. For example, if a consumer comes to a site, views several pages, and then buys a life preserver, who is he (or she?) An environmentalist going on a canoe trip? A powerboater? Or a hydrophobe? If a vendor tries to take any "clickstream-clever" action do something with this information, they will likely do more harm than good to their efforts. The attempt to profile the consumer will be wrong - very wrong - two of three times. A canoeist who is marketed to like a powerboater, or a hydrophobe offered more watersports gear, will likely be offended by the vendor. Practically, the vendor would not risk making such an assumption.
Clickstream profiling is too complicated to work.
Aggregating clickstreams from various sites and data mining a useable profile is too cumbersome a process. If the vendor attempted to aggregate information from the consumer's "online footprints" in order to profile him or her, the variety of possible "previous page-views and life jacket purchase" profiling rules rapidly overwhelm any system, much less a person trying to manage the web site. To effectively pinpoint a user profile, a web site's profile system would have to account for the number of pages, multiplied by the number of products, multiplied by the possible path through the site. In other words, a site with 100 products on 100 pages would have to take into account 100 x 100 x 100, or 1 million possible visitor patterns and then create a particular rule for each of the one million patterns. Obviously, this would not be practical.
Our company has recently been working with a vendor who has the most sophisticated anonymous clickstream retail system available. They have spent great resources implementing and managing the monolithic system, and they have recently reduced the reach and analysis of the system to where they will categorize their customers and visitors into one of only six different profiles. After a year of using the system and devoting enormous resources to it, they found that anything else is counterproductive.
Comment to Question # 6: "Are there technologies in development that will enable the creation of consumer profiles that identify the individual consumer?"
FollowUp.Net is a permission-based profiling company where the vendor asks their existing, known customers for profile input. Often this information is mixed with a consumer's clickstream record by matching cookie information from the consumer's computer. This profiling takes the form of an online survey sent to the customer, which the customer elects to complete. In some cases, the customer is incentivized to provide the information with a "discount-on-the-next-purchase" coupon. This information is then stored in a "Customer Profile Database" on our servers, with the information entirely owned by the vendor. This information is then used to target customers with particular e-mail marketing, provided the consumer has not opted-out.
Chris Woods is the founder and CEO of, FollowUp.Net, a company that provides a hosted, online customer profiling and target marketing service for retailers, manufacturers and service providers. We provide the software and host the customer data. Our clients include several of the largest and most innovative on-line retailers, as well as businesses whose primary distribution channel is "off-line."
FollowUp.Net's only develops and works with non-anonymous customer profiles. Our services specifically enable a vendor to profile his existing, know customers through direct surveys and customer-provided, opt-in responses. Our company believes that this is not only the most ethical way to collect customer information, but that customer-supplies information is much more useful marketing data than "click stream" analysis and cookie data mining.