COMPANY OVERVIEW

1
COMPANY OVERVIEW
2
ONLINE INSIGHT INC.

• Founded July 1998 by President/CEO Ken Forster
• Raised 250K in seed capital with COO Paul Krebs
• Developed Alpha products w/leading academics
• Raised 4.5M in first round venture from
  Greystone Capital
• Launched Precision Choice 1.0
• Put senior management and core team in place 50
  FTEs.
• Landed six customers, four financial services,
  two outside
• Partnerships with key integration, RD,
  technology players
• Launched Precision Choice 2.2, Precision Insights
  2.0

3
Online Insights solutions i) Provide the
foundation for online guided selling
environments WHILE AT THE SAME TIME ii) Capture,
analyze, and allow companies to act on
quantitative insight into customers buying
needs, preferences and motivations.
4
What is guided selling and why is it critical to
take e-commerce forward?

• Commerce sites are seeking to
• Raise close rates
• Provide high-touch, one-to-one selling to
  self-service customers (especially mainstream
  segments)
• Offer products that are complex, multi-feature,
  require greater consideration, involve trade-offs
• Current selling tools
• Fall short of delivering these capabilities
• Cater to experienced buyers
• Focus on marketing, not selling
• The answer is guided selling
• Consultative on-line sales processes that enable
  buyers to efficiently define needs and make
  decisions in a self-service environment.

5
Where guided selling fits
6
Guided sellings main elements
7
Sites must also monitor and interpret the their
online selling interactions.

• Why?
• Effectively manage the sales process itself
• Assess how buyers needs are met by current
  products, pricing, marketing campaigns, etc.
• Tracking buyer clickstream/mining sales
  transactions (i.e., who bought what) gives
  little understanding of motivations behind sales
  interactions.
• Dumb data plus black-box analytics
• Knowing how to ask the right, intelligent
  questions to build intelligent profiles is the
  key.

8
Precision Solutions is an integrated sales
environment management platform.
9
We are expanding our capabilities to fill a
larger market footprint.

• Facilitating purchase B2B/net marketplaces and
  strategic sourcing platforms
• Driving targeted marketing systems
• Incorporating additional algorithms and customer
  interaction approaches in Precision Choice
• Integrating with configurators for mass
  customization

10
Management team

• Ken Forster, President/CEO Company founder, 6
  years e-commerce/online fin. svcs. strategy
  development experience, Dartmouth graduate
• Paul Krebs, COO 5 years e-commerce /online
  fin. svcs. strategy development experience,
  Williams graduate
• James Sandry, CFO Founding officer of iXL
  Enterprises. During tenure at iXL he held the
  positions of CFO, Treasurer, and Executive Vice
  President
• Daryl Wehmeyer, Director of Analytics 10 years
  leading teams to develop user profiling and data
  mining models at IRI and Kelloggs
• Charles Flowers, VP Software Development 8
  years experience. Led development organizations
  at Mapics. Recognized Java thought-leader
• John Grendi, VP Sales Business Development
  Former Senior Category Marketing Manager at
  Dell Computer and VP Sales at Chase Manhattan
• Duane Cunningham, VP Professional Services 20
  years software implementation experience. Former
  Director at USWeb/CKS
• Scott Sanders, Director of Product Development
  6 years e-commerce experience. Former Director
  at iXL. Strategic Planning IBM.

11
Vertical market focus

• CURRENT
• Financial Services - Mutual funds, Mortgages,
  Credit Cards, Insurance

• ADDITIONAL VERTICALS
• Computers - Laptops, Desktops, Servers, Palm,
  Software, ISP services
• Consumer Electronics - DVD, Stereos, TVs,
  Camcorders, Cellular Services
• B2B Net Marketplaces, Small Business, Services
• Real Estate - Homes, Apartments, Condos,
  Timeshares
• Automotive - Cars, Trucks, SUVs, New/Used, Tires,
  Automotive Audio
• Travel - Airline Packages, Hotels, Destinations,
  Cruises
• Career Services - First Job, Career Change, Offer
  Negotiation
• Outdoor - Tents, Backpacks, Hiking Boots
• Lifestyle - Cities, Pets, Colleges, Health Clubs,
  Charities, Restaurants, Wine

12
Overview of Online Insight Business Issues

• Overview of Research Questions
• Two Fundamental Business Questions
• Usability of online recommendation technologies
• Credibility of online recommendation technologies
• Overview of Recommendation Technologies
• Six Online Guided-Selling Applications

13
Overview of Research Questions
Online Insight is interested in gauging consumer
feedback regarding our Precision Choice
recommendation technology and several competitive
decision tools.

• Question 1
• How do consumers view the usability of the
  various decision tools to be studied?

• What are consumers overall perceptions of the
  interaction with each technology?
• Was the technology easy to use?
• Did it require a significant time and effort
  investment?
• Was the tradeoff between the effort required and
  the accuracy obtained worthwhile?
• Do some tools require more knowledge on the part
  of the consumer to be used effectively?

14
Overview of Research Questions
Question 2 How do consumers characterize the
credibility of the various decision tools?

• Consumer-perceived accuracy of recommendation?
• Did the tool respond effectively to consumers by
  accurately identifying their needs?
• How much confidence do consumers place in each
  tools feedback?
• What factors influence consumers trust and
  confidence?
• What issues does each tool raise in terms of
  consumers willingness to provide the needed
  data? For example, are there privacy concerns
  associated with any of the tools?

15
Overview of Research Questions

• Compare and contrast the strengths and
  weaknesses of the various technologies on two
  dimensions.
• What types of product categories lend themselves
  to the various recommendation approaches?
• For example, how do consumers information needs
  differ for Computers, Financial Services, Autos,
  CDs, etc.?
• How does the traditional, offline sales process
  influence consumers perceptions of the online
  sales process for a product category?
• How does consumer feedback vary for different
  types of consumers?
• Price-sensitive or brand loyal shoppers versus
  shoppers focusing on other product features?
• How does a consumers experience-level with a
  particular category affect their feedback?

16
Overview of Recommendation Technologies
Technology Examples
Conjoint-based Precision Choice www.smallbizplanet.com
Constraint-based Yahoo Mutual Funds screen.yahoo.com/funds Personalogic (constraint plus) www.personalogic.com
Multi-Attribute Utility Theory Frictionless Commerce www.frictionless.com Purchase Source Retail Product
Configurator Dell Computers www.dell.com
Collaborative Filtering Amazon.com www.amazon.com
Artificial Intelligence Ask Jeeves www.etown.com (ask Ida)
None www.computer.com
17
Conjoint-based

• Conjoint analysis takes into account the fact
  that consumers make complex decisions based on
  several factors jointly rather than one factor at
  a time.
• Products or services are constructed in the form
  of product "profiles. Each profile is a
  combination of one selected level from each of
  the attributes.
• Attributes represent the key features of products
  or services
• Levels represent specific points along the key
  attribute dimensions.

Attribute
Levels
Brand
IBM, Compaq, Dell
Example Profile
IBM 600 mHz 2,000
Speed
500, 600, 700 mHz
Price
1,000, 1,500, 2,000
18
Conjoint-based

• Respondents are led through a two step process to
  obtain their preferences on key product
  attributes.
• First, respondents are asked to rate the
  importance and desirability of various product
  attributes.
• Second, they are presented with a series of
  product profile pairs. They are then asked to
  indicate a scaled preference to one of the
  profiles.
• Data produced by this process allows us to
  quantify respondents likes and dislikes, and
  determine the strength of preferences.

Customer Profile
1) User Rates Product Feature Importance and
Desirability
2) User Selects Most Preferred Products From
Hypothetical Pairs
Result Quantified Customer Profile
19
Constraint-based

• Constraint-based recommendation technologies
  utilize a database filtering methodology.
• The users set of potential product
  recommendations initially includes the entire set
  of products in the database.
• The user then sets certain specifications which
  enable products to be removed from the
  recommendation set (e.g. eliminate all products
  that cost more than 500 or only include
  products with a quality rating of at least 4
  stars.)
• As the user creates constraints, the set of
  available products diminishes. At any point, the
  user can view all products that are still
  available, and then choose to change previously
  set constraints or add new ones.
• Constraint tools are most appropriate for
  databases of fixed products, rather than
  build-to-order products.

20
Multi-Attribute Utility Theory/Stereotyping

• Multi-Attribute Utility Theory defines products
  by their features. Features are in turn defined
  by a number of attributes.
• For example, laptops might be defined by the
  following features Manufacturer, Memory,
  Operating System, Price, Processor, Screen Size,
  and Weight.
• The user specifies the range of acceptable values
  for each attribute within each feature.
• For example, for a numeric attribute such as
  Screen Size, the user could specify a range of
  13.3 to 15.1.
• The user then ranks the importance of each
  feature. Based on this information, the software
  scores each product in the product database.

21
Multi-Attribute Utility Theory/Stereotyping

• Stereotyping is a methodology used to expedite
  the preference profiling process.
• Users are given the ability to choose from one of
  several pre-established stereotypes which sets
  default preference ratings for each attribute.
  For example, for laptops, stereotypes might
  include business traveler, family user, and
  budget buyer.
• The preset default preference ratings would
  differ for a business traveler versus budget
  buyer and would emphasize different product
  attributes.
• Priority Features
• Business Traveler Budget Buyer
• Laptop weight Price
• Battery life Financing terms

22
Configurator

• Configurators enable customers to build
  made-to-order products while ensuring that
  incompatible features are not paired together.
• Configurators contain the rules and logic that
  dictate product assembly, ensuring that no
  product is configured that cant actually be
  assembled.
• To the user, configurators often appear very
  similar to constraint tools.

23
Collaborative Filtering

• Collaborative filtering is a methodology used for
  predicting a customers preferences for
  particular products based on the preferences of
  other users who have exhibited similar
  characteristics.
• It can rely either on explicitly entered
  preferences, observed click-stream behavior, or
  transaction history.
• Users are placed in segments and are served
  recommendations based on the observed preferences
  of others in the segment.
• From the users perspective, collaborative
  filtering is typically not an explicit process
  rather the user simply receives a recommendation
  without providing any explicit input.

24
Artificial Intelligence

• Artificial Intelligence-based recommendation
  technologies engage customers in a dynamic
  question-answer dialogue in order to uncover
  their needs and preferences.
• Questions can be in natural language form or in
  multiple choice form with radial buttons and
  check boxes.
• Typically, the questions pertain to how the user
  plans to use the product.
• Also, some constraint-type questions are usually
  included. Based on the answers to the questions,
  the technology assigns weights to various
  attributes and scores the products in the
  database.