Transcript Merchant Systems - UCL Computer Science

Intelligent Marketing
Dr Tim King
22nd November 2006
My CV

Cambridge Computer Lab 1973-1981
– Wrote a relational database for Ph.D.


Lecturer, University of Bath 1981-1983
R&D Director 1984-1986
– Wrote AmigaDOS

Founded Perihelion 1986
– Distributed OS, embedded systems, database systems

Founded UK Online 1994
– First UK ISP with content
– Sold to EasyNet 1996

Independent consultant
– Technical Due Diligence for VCs
– Advice for Sainsbury’s, Sony, Home Office etc
– Strategy for small companies and following M&As
Intelligent Marketing
Using computers to make intelligent use of easily available data

Market drivers
–
–
–
–

Supply and demand
One size does not fit all
Cost curves
Price sustainability
Marketing techniques
–
–
–
–
–
Lock-in and Lock-out
Brands
Up sell
Personalisation
Customer support
Basic Economics
Price p
Demand Curve
D(p)
Supply Curve
S(p)
p*
People pay more for something rare
Mass production reduces cost
Quantity
Market fit


Different versions for
different market subsections
Many examples
Price p
– Travel
• First class vs coach
– Cars
• Audi vs Skoda
– Software
• “lite” versions with hardware

Psychology
– Brand awareness
Quantity
Cost Curves
Cost p
Cost p
Price p
Price p
Quantity
“General Motors”
Quantity
“Microsoft”
Price sustainability
Price of iPod 60G
Time Period: 20 Dec 2004 to 31 Oct 2005
Red = Highest , Blue = Average, Green=Lowest
Lock-in

Buying something commits you to buying more
– Services
– Consumables
– Complimentary products

Examples:
–
–
–
–
–
Operating systems
Ink-jet printers
Mobile phone subsidy
Car services
Frequent flyer
Lock-out



Incumbent tries to maximise switching cost
Loyalty programs
Technology control
– Nintendo game cartridges
– Sony Playstation DVD formats


Crypto and tamper resistance
Community – its where your friends are
• BB, chat for registered users

Hassle
– eg email address change
Brand awareness


Single most important piece of data
People buy from a known name
– Sense of trust
• Marks and Spencer
– Perceived value
• Cheap reliable airline => Cheap reliable mobile
– Peer pressure
• Nike, Rolex, Dolce and Gabanna, Ferrari

Brands can expand
– Virgin
• Active, Atlantic, Books, Brides, Broadband, Cosmetics, Credit
card, Drinks, Galactic(!), Games, Holidays, Limobike, Megastore,
Mobile, Trains, Wines
– Apple
• From computers to iPods
Expand the Brand (1)

YouTube
– TV adverts
• Recycle TV adverts
• People send copies of your advert to each other
• Risqué adverts not acceptable on TV
– TV shows
• Trail shows
• Repeat the best bits
– Music
• Shareable
• Do-it-yourself MTV
Expand the Brand (2)

Google
– Buy your brand name
• Coke
–
–
–
–
Careers
Corporate Responsibility
The Coca-Cola company
Press Centre
– Buy your supplier’s brand name
• Nike
– JDSports
– Buy your competitors’ brand name
• Ford
– Adverts for Seat dealer
– Buy your target
• Nike (Boycott Nike), Coke (KillerCoke)
Up Sell

Persuade people to buy more
– Buy two get one free
• When you only wanted one in the first place
– Packs of three
• When you only want two batteries

Persuade people to buy something else
– Dell
• Insist that you take extra warranty
– Travel sites
• Insist that you buy travel insurance
Personalisation (1)

Know your customer
– Profile typical users when they visit a web site
• Purchase history
• Time to make purchase decision
• Amount of research done
– Profile users through loyalty cards
• Nectar
– They know everything you have ever bought
– Keep in touch with customers
• Email newsletters
– Lastminute, Maplin
• Cookies
– Welcome back Tim
Personalisation (2)

Know your customer type
– User database
• Address/postcode -> socio economic indicator
• Gender
• Age -> Register with Data Protection Registrar
• 60 “bins” 5 classes x 2 genders x 6 age groups
– (kids, teens, dinky, married with kids, empty nesters, retired)
– Disposable income
– Disposable leisure time
– Recommendation
• People who bought this also bought that
– Data from your own site
– Amazon really can recommend music or books you might like
• Data mining
– People who buy this on cold winter Fridays in Slough also buy that
Customer support

Identify meaning of email
– Auto-respond with the answer

Classify once human response given
– So next time it will auto-respond

Expose database as FAQ
– So they don’t send the email at all

Always give the option of human interaction