Transcript Introduction - Homepages | The University of Aberdeen

E-Commerce – customer focus

Attracting and keeping customers
» Issue: security, trust
Legal issues
 Personalisation
 Adverts

Dr. Ehud Reiter, Computing Science, University of Aberdeen
1
Customers are different!

Consumer types
» Individual consumers
» Organizational buyers

Goal of shopping
» Pragmatic: buy something useful, cheaply
» Hedonic: have fun

Personality
» Impulsive buyers—purchase quickly
» Patient buyers—make some comparisons first
» Analytical buyers—do substantial research before buying
Dr. Ehud Reiter, Computing Science, University of Aberdeen
2
Consumer
Behaviour
Dr. Ehud Reiter, Computing Science, University of Aberdeen
3
Prentice Hall, 2002
Consumer Satisfaction
Dr. Ehud Reiter, Computing Science, University of Aberdeen Prentice Hall, 2002
4
Security/Trust

Security/trust
» Will the company actually deliver the
correct product in reasonable shape, in a
reasonable time, at correct price
» Will the customer pay up (is the credit card
stolen, will it be repudiated)
Technical aspects
 Human aspects: Focus here on trust

Dr. Ehud Reiter, Computing Science, University of Aberdeen
5
Trust in physical shops
Experience: shoppers trust shops
they’ve used before
 Appearance: shoppers trust store that
look reputable
 Complaints: easy to complain, shop
can’t hide
 Transactions are simple

Dr. Ehud Reiter, Computing Science, University of Aberdeen
6
On-line trust

What makes you trust an e-commerce
shop?
Dr. Ehud Reiter, Computing Science, University of Aberdeen
7
On-line Trust

Experience: I trust Amazon because I’ve
used them before
» Reputation: because my friends use them
» Very important with e-shops

Appearance: do I trust Amazon because
they have a nice website???
» Less important than with physical shops
» Marketing helps
Dr. Ehud Reiter, Computing Science, University of Aberdeen
8
On-line trust
Complaints: Harder to complain since
don’t know where shop is
 Transactions are complex because of
delivery

» Where many e-shops mess up

Third-party: do I trust Amazon more if
another web site says good things
about Amazon (???)
Dr. Ehud Reiter, Computing Science, University of Aberdeen
9
Does Amazon Trust Me?

Amazon trusts me because
» Experience: I’ve always paid Amazon
before
» Reputation: I’ve used other companies and
always paid up
» Marketing: Amazon threatens nasty things
to customers who don’t pay up
Dr. Ehud Reiter, Computing Science, University of Aberdeen
10
Trust
We know how trust is established in
physical shops.
 We are developing mechanisms for
establishing trust in e-shops

» Partially technology, but psychology and
sociology probably matter more
» Lack of trust mechanisms is barrier to new
e-shops
Dr. Ehud Reiter, Computing Science, University of Aberdeen
11
Legal Issues: Tax

In USA, one driving force behind early
e-store success was less tax
» Because of a tax loophole, sales tax (VAT)
was not charged on e-commerce sales

Automatically gave price advantage to
e-commerce sites!
Dr. Ehud Reiter, Computing Science, University of Aberdeen
12
Legal Issues: Intl E-Commerce
In theory, e-commerce means sites can
sell globally
 In practice, difficult because of different
tax rules, regulations, customs, etc

» More common to set up subsidiaries in
different countries, as Amazon has done

Lack of global legal/regulatory
framework hinders ecommerce
Dr. Ehud Reiter, Computing Science, University of Aberdeen
13
Personalisation

E-Commerce sites can treat customers
differently
» Offer recommendations, special deals
» Personalise web site
» Adjust prices

In theory, “personalised shop” one of
the great benefits of e-commerce
Dr. Ehud Reiter, Computing Science, University of Aberdeen
14
One-to-One Marketing
Build a long term association
Meeting customers cognitive needs
 Customer may have novice, intermediate or expert skill
E-loyalty—customer’s loyalty to an e-tailer
 costs Amazon $15 to acquire a new customer
 costs Amazon $2 to $4 to keep an existing customer
Trust in EC
 Deterrence-based —threat of punishment
 Knowledge-based —reputation
 Identification-based —empathy and common values
 Referrals – Viral Marketing
Personalisation…
Personalisation - Marketing Model
“Treat different customers differently”
Prentice Hall, 2002
Personalisation
“Process of matching content, services, or
products to individuals’ preferences”
Build profiles – N.B. Privacy Issues
 Solicit information from users
 Use cookies to observe online behavior
 Use data or Web mining
Recommendation

Build profiles
» What has X bought?
» What has X looked at?
» Demographics: age, gender, etc

Recommendation
» Rules: If X buys Harry Potter 6, recommend HP 7
» Data Mining: Other people who bought Harry
Potter also bought Lord of the Rings
» Collaborative: X’s overall buying profile is similar
to Y, so recommend whatever Y bought
Dr. Ehud Reiter, Computing Science, University of Aberdeen
18
Data Mining
searching for valuable information in extremely large databases
Automated prediction of trends and behaviors
 Example: from data on past promotional mailings, find out
targets most likely to respond in future
Automated discovery of previously unknown patterns
 Example: find seemingly unrelated products often purchased
together
 Example: Find anomalous data representing data entry errors
Mining tools:
 Neural computing
 Intelligent agents
 Association analysis - statistical rules
Web Mining - Mining meaningful patterns from Web resources
 Web content mining – searching Web documents
 Web usage mining – searching Web access logs
Recommendations

If done well, perceived very positively
» Real benefit, not just marketing spam
» Credit-card companies have done this well
– Have the most purchasing data?

Data privacy issues
» Can Visa sell data about you to Amazon?
» Spyware to track all of your web browsing?
Dr. Ehud Reiter, Computing Science, University of Aberdeen
20
Personalise Web Sites
Let customers create their own “shop
front” focusing on their interest
 Adjust appearance (eg, for visually
disabled, or strict Muslims)
 Doable, not huge success

Dr. Ehud Reiter, Computing Science, University of Aberdeen
21
Personalised Pricing

Companies would love to be able to
charge people different amounts for the
same product
» Airline seats, cars, etc
» Full price for people who are keen, in a
rush, don’t care about money
» Discount for choosy/finicky
Dr. Ehud Reiter, Computing Science, University of Aberdeen
22
Personalised Pricing
Amazon, etc have tried this, but
customers hated it.
 So has gone “underground” for now.
 Technology permits this, but society’s
expectations does not allow it

Dr. Ehud Reiter, Computing Science, University of Aberdeen
23
Advertising

E-Shops (and other sites) can make
money via advertising
» Google makes billions from its “sponsored
links”
» Amazon has adverts as well
Dr. Ehud Reiter, Computing Science, University of Aberdeen
24
Web Advertising
Conventional advertising focuses on
visual appeal
 Less successful on web

» Flashy animated banner adverts are a
nuisance and distraction
Dr. Ehud Reiter, Computing Science, University of Aberdeen
25
Targeted adverts

Web allows relevant adverts to be
associated with a web page
» Google sponsored links based on search
» Amazon could display different adverts for
sci-fi and romance novel

Very effective if done well
» So Web sites can charge more for targeted
adverts
Dr. Ehud Reiter, Computing Science, University of Aberdeen
26
Web adverts
Initially treated like TV adverts, put huge
effort into flashy multimedia banner ads
 Now focusing on simple targeted
adverts instead
 Advertising models cannot be blindly
moved from TV to web

» need new models!
Dr. Ehud Reiter, Computing Science, University of Aberdeen
27
E-Commerce Summary

Initially tried to make e-shops similar to
high street shops. But
» Need different business model
» Trust issues much more important
» Need appropriate legal framework
Dr. Ehud Reiter, Computing Science, University of Aberdeen
28
E-Commerce Summary

Sometimes technology really helps
» Recommender systems, targeted adverts

Sometimes technology works but
society doesn’t like it
» Differential pricing
Dr. Ehud Reiter, Computing Science, University of Aberdeen
29