AI is a misnomer. Implying sentience leads to over hyped expectations.
Ian talks to the reality of AI application in marketing and startups.
He focusing on what AI is being used for now, hints at its role in the future and reflects on his own “AI” projects over the years.
Frameworks for adoption are presented with some clues as to who to work with and how to work with vendors for AI success.
2. Ian Morrison
AI in Marketing
Marketing industry veteran with a background as agency CIO, Product Director and Creative
Technologist.
Founder & CEO of GKIM Vietnam, a 25 man boutique dev shop delivering work for
• Agencies such as Ogilvy
• The Singapore Government
• Ambitious entrepreneurs around the world
An international career with many tech achievements
• AI algorithms for popular video games
• Designed chatbots pretending to be God
• Applied AI to many digital business solutions
5. AI is a Misnomer
Intelligence is the domain of Sentient Beings.
AI is Data Science, Algorithms and Smarter UX’s.
A catch all term for smart technology and solutions.
6. Different Things To Different People
Maths & Data Science
• Exploring mazes of data
• Seeking seeking
In Games
• A few lines of code
• Entertain for hours
In Business
• Optimize Supply and Demand
• Configure the Marketing Mix
7. For Marketers
2. New Human Computer Interfaces (HCI)
Cognitive AI’s enrich communications
Customer Service apps
Engage Richly with Human Brain
Driven by Technological Capability. Two Broad Application areas.
1. Smarter, Data Driven, Digital Marketing
Analytics, Machine Learning & Expert Systems
Smarter marketers
Engage Smarter using Computer Brain
8. #1 Smarter Digital Marketing
Given
a) clearly defined goals
b) enough data to analyze.
c) enough parameters to play with
AI will find optimal solutions for the
marketing mix.
Engage
Smarter
Price
Quantity
9. #2 New HCI’s
Interactive Messaging, Customer Service & Sales Bots
Face analysis*, Speech
recognition*, NLP*
NLG*, text to
speech*, 3D Faces*
Human
Communications
Cognitive
Tech
Programmed
Brain*
Words & Sentences
Voice & Tone
Body Language
Digital CX Manager
Virtual Girlfriend
Robot Doctor
Engage
Richly
Provided by multiple
competing vendors
Chat Bot
Creative Opportunity
Developing real personalities.
Brand personas become actual personas.
Apps named and treated as people.
11. AI and the 4 P’s
Product
Recommendation
Creative
Copywriting
and Content
Individual, Time
based Pricing
Digital Media
Targeting
12. Product
Recommendations - TV Shows, Music & Marketplaces
Premise
Customers don’t know what they want until they are told
A strategic part of buying cycle
Lists of crap - put you right off
Have to go find what you want – not bother
Given something that delights - no hesitation
Torrent services miss this convenience & losing favor
In the business of rights ownership and marketing
AI recommendations worth $1b to Netflix
AI exploits the long tail with infinite customization
13. Price
Optimizing ARPU - airline tickets, hotel bookings and taxi fares
Pricing Decisions
Personal, Time sensitive
Strategic & Complicated
Commercialize inventory
Maximize long term ARPU
Game the competition
Why travel?
Fixed supply
Variable demand
Elastic pricing
Night time Taxi Demand
Personalized Pricing
14. Place
Media Optimization - What digital advertising is all about
Goals of digital advertising
Maximize media opportunity
Targeting and retargeting
Competitive algorithmic buying
Strategic advantage from AI
Looking for patterns of browsing, clicking and purchase
Discovering new opportunities, market segments and successful content
Anticipating consumer behaviour
Requires Scale
Data warehouses
AI power
Fastest networks
15. Promotion
Algorithmically Generated Content & Copy - Curation & Creation
1. Customized online recommendations and calls to action’
Trivial, fun and powerful
2. Narratives in focused domains
Rules laced with content variations and NLG
Educational content
Sports narratives
GKIM working on AI assisted content generation for PR
3. Advertising copy
Human empathy required!
Supporting different demo, psycho and geo..
Benefits must
justify complexity
and cost
16. Typical Customer, MAXPLORA
Problem
Rich database of deals
Total value $12b
Complex buyer and seller profiles
Need
Accurate recommendations
Suggested buyer / seller matches
SQL doesn’t cut it any more
Solution
Customized Amazon Elastic Search
scores and ranks deals
Result
Powerful recommendations
Very cheap technology
A “poor man’s AI”?
Pareto says “job well done”!
Many applications don’t need “AI” per se
17. Giving Voice to God & other Brands
Branded Conversations
• Infer meaning from user texts
• Develop libraries of responses
• Constantly tweak scoring, manually
• “What do you think?”, God would say if he got confused
God’s AI
• State machines and conversation trees
• Tech for dozens of campaigns
AI designed to support the Creative experience
20k bots deployed. Alexa & Siri are the HCI of our kids.
18. AI Bot Research Objectives
The richest of possible user interfaces
• Text
• Audio
• 3D Computer Graphic Avatars
The smartest of AI brains with
• Ability to maintain a conversation
• Knowledge base (content) to engage
• Drive consumers around buying cycle
Next Generation Chat Bots
19. Does Marcel pass the Sniff Test?
Fuzzy and complicated domains?
• Creativity
• Collaboration
• Marketing solutions
All in one little box?
Natural language voice interface?
June 2018 exactly?
Marcel as a proxy for GKIM’s own research
Leave tech sales to tech people?
20. But Wait...
Context is controlled
Question are straightforward - who? what?
Database exists - people and projects
Face analysis,
Speech
recognition,
NLP
NLG, text
to speech,
CG
AI is available now
7 areas of AI
Commoditized already
Conclusion
Advertising over hyped
Product could be real
Is voice the right UI?
22. Build your own AI?
Challenges
• Google Goggles face processing – inconsistent results with low CPU
• Super high tech 3D Painting – one week became six months
• Live actor driven avatar technology – should have just licensed
Characteristics
• Technology must serve
• Is very difficult
• Available very cheap
High Tech Project Experience
3rd Party AI’s and frameworks
are the only option
23. Marketing clouds
• Major vendors have AI offerings
• Branded toolsets
• Adding spice to their CX, CRM and marketing tools.
• Not so much generic AI.
Core Tech vendors
• Rich suite of AI API’s
• Interchangeable
• Getting better all the time
Who to work with
Core Tech or Marketing Tech
24. Working with Marketing Vendors
AI should
● Add value to some components
● Develop insights from data pool
● Automate marketing process
Client Businesses
Integrating and Deploying the tailored
marketing solution
AIMarketingStakeholderStack
Agency’s AI Marketing Solutions
Industry specific marketing, AI personas,
Content & Tech
Data Science Provider
AI algorithms, historical data and training
models
Vendor’s Marketing Clouds
Kotler’s AI brain drives marketing
automation.
AI enhanced Digital marketing tools
Complicated?
25. Working with App Developers
Customer Touch Point
Delivering the smarter UX
Marketing Driven UX Tactics
Application specific
programming and integration
Data Science
AI algorithms, historical data
and training models
AI should
● Show its face to consumers
● Support marketing tactics around the UX cycle
● Control all aspects of the UX
AIAppStack
Fun!
27. AI Marketers will Emerge
Adopt Technology
Accessible and Interoperable Technologies
New, AI driven marketing clouds
Learn Skills
Leave the AI tech to others
Develop AI marketing expertise
Takes time for an Industry to Change
28. AI Momentum is Unstoppable
Cynical for now
Ignore the name
It’s wrong, misleading and distracts
from what it really is
Be nervous
Big promises and big prices
Ecstatic for future
AI is the magic ingredient
The power of code, data and connected systems
Making computers smarter, more than tools
Prepare to be delighted
When you can
• Ask AI the right questions
• Pick the right AI technology
• Design the right architecture
• …with a genuine AI personality at the core