Revenue Opportunities in the Voice Enabled Web

1
Revenue Opportunities in the Voice Enabled Web

• Bryan Duggan
• MSc Computing Science
• (Information Technology for Strategic Management)
• Advanced Research Module

2
What this paper is about?

• Introduction
• Definitions
• The voice enabled web
• Vcommerce
• A technology view
• VoiceXML
• Infrastructure
• A business view
• Revenue models
• Successes and failures
• BOSS model - Business Opportunities for Speech
  Systems

3
Definitions

• The Voice Enabled Web is any speech-enabled data
  interaction that utilises some browsing mechanism
  to navigate between separate sites including
  voice-enabled web sites and interactive voice
  response (IVR) systems.
• - The Yankee Group
• Vcommerce is an emerging term which describes
  the usage of speech technologies, (speech
  recognition, text-to speech and speaker
  verification), typically over the telephone in
  commercial applications such as banking, buying
  cinema tickets or stock trading - The
  Yankee Group

4
The Vcommerce Formula
5
VoiceXML

• To the speech enabled web what HTML is to the
  browser centric web
• eXtensible Mark-up Language
• W3C
• Voice XML
• audio files to be played
• text to be spoken
• speech to be recorded or recognised
• touch tone input to be collected
• HTML
• text
• graphics
• tables
• Hosted on a web server
• Dynamically or statically generated

6
VoiceXML

• Dialog
• Interaction between a user and the voice browser
• Input and output
• Grammar files
• Telephony
• Disconnect or transfer a call
• Call trobmoning
• Platform
• Platform specific features
• Performance
• Pre-fetch
• Cache
• dead air elimination

7
VoiceXML

• lt?xml version"1.0" encoding"Cp1252"?gtlt!DOCTYPE
  vxml PUBLIC '-//Nuance/DTD VoiceXML 1.0//EN'
  'http//voicexml.nuance.com/dtd/nuancevoicexml-1-2
  .dtd'gt
• ltvxml version"1.0"gt ltmeta name"Generator"
  content"V-Builder 1.2.30" /gt
• ltform id"intro"gt
• ltblock name"block1"gt
• ltprompt bargein"true"gtHello. This is your
  computer.lt/promptgt
• ltgoto next"mainMenu" /gt
• lt/blockgt
• lt/formgt
• ltmenu id"mainMenu"gt
• ltprompt bargein"true"gtWould you like
• ltenumerate /gt
• lt/promptgt
• ltchoice next"banking"gtbanking orlt/choicegt
• ltchoice next"weather"gtweatherlt/choicegt
• lthelpgtJust say banking or weather or sports
  and I will tell you the informationlt/helpgt
• ltnoinputgtHey, did you say something? I didn't
  hear. Try again.lt/noinputgt
• ltnomatchgtForgive me, I am a stupid computer
  and I didn't understand what you said. Please try
  again.lt/nomatchgt
• lt/menugt

8
Why is it important?

Enables creation of new services
Common development standard
Universal vendor acceptance
Why is VoiceXML important?
Encourages third party development
Leverages web resources
Lowers development costs
- Strachmann 2000
9
Ecommerce Infrastructure
Tier 1 User interacts with a web browser (HTML
and JavaScript)
Tier 2 Presentation tier, can be implemented in
ASP, JSP, Perl, PHP or any web scripting language
Application Server
Tier 3 Business Rules, can be implemented as
EJB, COM, CORBA or DCOM objects
Tier 4 Database, holds data which drives the
application
Multi-tiered web application model (Allamaraju et
al., 2001 Amended)
10
Tier 1 User dials in to a voice enabled web site
using a standard telephone
The call is carried over standard analogue
telephone connection
Tier 2 The call is answered by a computer
running a voice browser which understands
VoiceXML. The voice browser has built in speech
recognition and a text-to-speech engine
Vcommerce Infrastructure
Tier 3 Presentation tier, can be implemented in
ASP, JSP, Perl PHP or any web scripting language.
Generates VoiceXML instead of HTML
Application Server
Tier 4 Business Rules, can be implemented as
EJB, COM, CORBA or DCOM objects
- Harmen, 2000 General Magic Corp, 2002 Amended
Tier 5 Database, holds data which drives the
application
11
Conclusions

• VoiceXML
• 4 tier model for ecommerce
• 5 tier model for vcommerce
• The multi-modal web

12
Revenue Models

• Several revenue models have been tried, some
  successfully and some unsuccessfully
• Like web portols advertising driven, now seeking
  alternatives
• Voice portols
• Content from a number of sources is aggregated
• Email, news, stock quotes, weather reports,
  traffic news, restaurant recommendations, cinema
  reviews
• Low revenues
• Companies are increasingly turning to corporate
  voice portals

13
Revenue Models

• Advertising
• Like banner ads only more intrusive!
• 10 second ads before main menu
• Unpopular
• Linked or targeted advertising
• Fee based
• User pays a fixed fee to access the service
• More successful than the advertising model
• AOLByPhone 4.95 per month
• Quest Wireless twice the penetration of its web
  portol
• Talking Email 70 per year

14
Revenue Models

• Premium Rate
• Call is charged at a higher than normal rate
  (E.g. 1550 or directory enquiries service)
• Customer pays for the service on a per call basis
• Half goes to telecomm company, half to service
  provider
• Limited appeal for a voice portal
• Needs to be a killer app
• Games, customised ringtones and logos
• Only models which will work in a small market
  such as Ireland
• High cost of developing voice enabled web
  applications.

15
Revenue Models

• Phone Sales Model
• Sell items over the phone, no human intervention
• Ties in with a web based sales interface to
  record names addresses etc
• Office Depot Inc
• Place orders
• Check order status
• Browse the Catalog
• in place since October 2000
• 5 per cent of Office Depot's retail catalogue
  orders
• Orders cost 88 per cent less to process
• Refferrals
• Call tromboning
• Little success

16
Revenue Models

• Development and Outsourcing Model
• A company develops and hosts a voice application
  for another company
• Charges a hosting or development fee for the
  application or
• Charges a commission on transactions completed
  through the application
• A minimum of 250,000 for hardware and software
• A company exposes some of its back end
  infrastructure to an externally hosted voice
  application
• Takes advantage of existing infrastructure (See
  model)
• Analysts estimates that the market will be as
  much as 38 billion by 2007
• Content Provider Model
• Existing radio or television station
• Content management company

17
Revenue Models
18
BOSS Model

• Business Opportunities for Speech Systems
• A combined business and technology model for the
  voice enabled web
• Illustrates where in the voice enabled web,
  revenue opportunities exist
• Catagorises Revenue Models by tier

19
Referral Model - Businesses are charged for
referrals from a voice application
Fee Based Model - User is charged an up-front fee
to use the service
Tier 1 User dials in to a voice enabled web site
using a standard telephone
Premium Rate Model - Revenue is collected by the
teleco and shared with the vcommerce company
The call is carried over standard analogue
telephone connection
Tier 2 The call is answered by a computer
running a voice browser which understands
VoiceXML. The voice browser has built in speech
recognition and a text-to-speech engine
BOSS Model
Tier 3 Presentation Tier, can be implemented in
ASP, JSP, Perl PHP or any web scripting language.
Generates VoiceXML instead of HTML
Selling over the Phone Model - Back end ecommerce
infrastructures are reused to facilitate vcommerce
Tier 4 Business Rules, can be implemented as
EJB, COM, CORBA or DCOM objects
BOSS Model - Business Opportunities for Speech
Systems (Duggan, 2002)
Content Provision Model - Content providers
charge for customised content
Tier 5 Database, holds data which drives the
application
20
BOSS Model
21
Conclusions

• Voice Enabled Web
• VoiceXML
• Voice Infrastructure
• Revenue Opportunities
• Advertising
• Fee based
• Selling over the phone
• Referrals
• Development and Outsourcing
• Content provider
• BOSS Model
• Combines technology model with revenue
  opportunities

22
Questions?