This document provides an introduction and overview of artificial intelligence for marketing. It discusses how machine learning works using supervised, unsupervised, and reinforcement learning methods. It also outlines tips for bringing AI into a marketing organization, including clearly defining goals, having clean and consistent data, and starting with repetitive tasks like ranking, sorting, and anomaly detection. The document emphasizes that business stakeholders should use their industry and market knowledge to set goals, identify problems, and make decisions.
3. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
IS NOT
for data scientists
How to be a data scientist
Cool algorithms
Statistical tips and tricks
How to build a bot
Latest start-ups
Quantum computing
Intro to AI & Machine Learning
How some methods differ
How to talk to data scientists
Where AI/ML is used in marketing
How to bring AI/ML into your org
How to keep your job
IS
for marketers
4. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
Why Artificial Intelligence Now?
50 years of study
Cheap storage of Big Data
Massively parallel processing (GPUs)
Open Source
5. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
What is it?
How does it work?
What is it good at?
How is it useful?
Intro to Artificial Intelligence
for Marketing
6. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
As Seen on TV
AI: Anything computers can’t
SciFi: Anything AI can’t
“General AI” – thinks and acts human
Sentience
“Narrow AI” – task specific
Functional
7. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
Artificial Intelligence
Natural Language Processing
Speech recognition
Speech to text
Text to meaning
Sentiment analysis
"Wreck a nice beach"
14. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
Software Grows Up
Specific Logic Mathematical Model
Do this, then this, then this
If this happens, do that
If confused, report error
Statistical Model Machine Learning
{A:13,BUTTON:0,CHECKBOX:32,COMBOBOX:13,GRIDCELL:13,LINK:13,LISTBOX:13,MENU:0,MENUBAR:0,MENUITEM:0,MENUITEMCHEC
KBOX:0,MENUITEMRADIO:0,OPTION:0,RADIO:32,RADIOGROUP:32,RESET:0,SUBMIT:0,SWITCH:32,TAB:0,TREE:13,TREEITEM:13},G
=function(a){return(a.getAttribute("type")||a.tagName).toUpperCase()in ba},H=function(a){return
(a.getAttribute("type")||a.tagName).toUpperCase()in ca},ba={CHECKBOX:!0,OPTION:!0,RADIO:!0},ca={COLOR:!0,
DATE:!0,DATETIME:!0,"DATETIME-LOCAL":!0,EMAIL:!0,MONTH:!0,NUMBER:!0,PASSWORD:!0,RANGE:!0,SEARCH:!0,TEL:!0,
TEXT:!0,TEXTAREA:!0,TIME:!0,URL:!0,WEEK:!0},da={A:!0,AREA:!0,BUTTON:!0,DIALOG:!0,IMG:!0,INPUT:!0,LINK:!0,MENU:
!0,OPTGROUP:!0,OPTION:!0,PROGRESS:!0,SELECT:!0,TEXTAREA:!0};var I=function(){this.i=this.g=null}
16. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
Software Grows Up
Specific Logic Mathematical Model
Do this, then this, then this
If this happens, do that
If confused, report error
Statistical Model Machine Learning
17. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
Software Grows Up
Specific Logic Mathematical Model
Do this, then this, then this Describe numerical relationships
If this happens, do that Calculate alternatives
If confused, report error Human compares results & iterates
Statistical Model Machine Learning
18. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
Software Grows Up
Specific Logic Mathematical Model
Do this, then this, then this Describe numerical relationships
If this happens, do that Calculate alternatives
If confused, report error Human compares results & iterates
Statistical Model Machine Learning
Calculate probabilities
Project likelihoods
Human compares & iterates
19. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
Software Grows Up
Specific Logic Mathematical Model
Do this, then this, then this Describe numerical relationships
If this happens, do that Calculate alternatives
If confused, report error Human compares results & iterates
Statistical Model Machine Learning
Calculate probabilities Uses examples to figure it out
Project likelihoods and changes it's mind
Human compares & iterates
21. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
What is it?
How does it work?
What is it good at?
How is it useful?
Intro to Artificial Intelligence
for Marketing
23. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
Machine Learning
Supervised
You know the right answer
needs many examples
of labeled data
Dog: Yes Cat: No Tag a friend?
24. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
Machine Learning
Supervised
cat cat
cat cat
You know the right answer
needs many examples
of labeled data
25. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
Machine Learning
Supervised
You know the right answer
needs many examples
of labeled data
26. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
Machine Learning
Supervised
You know the right answer
needs many examples
of labeled data
27. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
Machine Learning
Supervised
You know the right answer
needs many examples
of labeled data
28. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
Machine Learning
Supervised
You know the right answer
needs many examples
of labeled data
29. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
Machine Learning
Unsupervised
Online buying goes up when the weather is bad
Ice cream causes drowning
Shoes cause headaches
Nicolas Cage is a monster
You don't know the right answer
Machine finds patterns in unlabeled data
may or may not be useful (correlation/causation)
31. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
Machine Learning
Supervised
You know the right answer
needs many examples
of labeled data
Unsupervised
You don't know the right answer
finds patterns in unlabeled data
may or may not be useful (correlation/causation)
32. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
Machine Learning
Reinforcement
There is no absolute right answer
some answers are better
"rewards" based on results
optimizes over time
Photo by Marek Szturc on Unsplash
33. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
3 Needs for 3 Deeds
of Machine Learning
Data Detect
Goal Decide
Control Revise
Needs Deeds
34. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
What is it?
How does it work?
What's it good at?
How is it useful?
Intro to Artificial Intelligence
for Marketing
35. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
Learning Machine Learning
High Dimensionality
High Cardinality
What's it good at?
36. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
Dimensionality = Attributes per Object
Cardinality = Options per Attribute
37. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
Dimensionality = Attributes per Object
Cardinality = Options per Attribute
Person
Objects
Attributes
billions of permutations
38. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
What is it?
How does it work?
What's it good at?
How is it useful?
Intro to Artificial Intelligence
for Marketing
47. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
Artificial Intelligence
Natural Language
Speech to text
Conversation Bots
Text to meaning
51. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
Artificial Intelligence
Natural Language
Speech to text
This means that
Repeated correction
Taught over time
Conversation Bots
Text to meaning
Concept & emotion imitation
Repeated correction
Taught over time
Lenovo Unified
Customer Intelligence
53. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
Artificial Intelligence
Natural Language
Speech to text
This means that
Repeated correction
Taught over time
Conversation Bots
Text to meaning
Concept & emotion imitation
Repeated correction
Taught over time
Lenovo Unified
Customer Intelligence
54. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
Artificial Intelligence
Natural Language
Speech to text
This means that
Repeated correction
Taught over time
Conversation Bots
Text to meaning
Concept & emotion imitation
Repeated correction
Taught over time
56. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
Bringing AI Into Your Organization
Look what followed me home!
Can we keep him?
62. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
AI Onboarding Tips
Clearly identified goals
Know Your Data
Would I advise my uncle?
Would I stake my reputation?
Would I risk my own money?
Would I bet my job?
63. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
AI Onboarding Tips
Clearly identified goals
Know Your Data Clean, Consistent, Reliable
64. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
AI Onboarding Tips
Clearly identified goals
Know Your Data Clean, Consistent, Reliable
Server Outage
Missing Tag
Broken Tag
Corrupted Data
Broken ETL
Enrichment Error
Integration Error
Ad Blocker
Etc., etc....
65. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
AI Onboarding Tips
Clearly identified goals
Know Your Data Clean, Consistent, Reliable
Web Analytics, Customer Service, Customer Relationship Management, Sales Force Automation,
Campaign Analysis, Social Media, Email Marketing, Mobile, Wearables, Surveys, Facebook,
Aggregators, App Data, Behavioral Data, Accounting, Video Views, API's, Enrichment, etc., etc....
66. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
AI Onboarding Tips
Clearly identified goals
Know Your Data
Web Analytics, Customer Service, Customer Relationship Management, Sales Force Automation,
Campaign Analysis, Social Media, Email Marketing, Mobile, Wearables, Surveys, Facebook,
Aggregators, App Data, Behavioral Data, Accounting, Video Views, API's, Enrichment, etc., etc....
Clean, Consistent, Reliable
67. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
AI Onboarding Tips
Clearly identified goals
Know Your Data Clean, Consistent, Reliable
68. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
AI Onboarding Tips
Clearly identified goals
Know Your Data
Data Lake
Clean, Consistent, Reliable
69. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
AI Onboarding Tips
Clearly identified goals
Know Your Data
Data Lake
Clean, Consistent, Reliable
73. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
Correlations people with this attribute have that attribute
Segmentation these people form a group
Clustering there are X number of groups
Anomalies these people are unique
Are results interesting? Useful? Worthy of further study?
What Can ML Do Better?
74. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
AI Onboarding Tips
Clearly identified goals
Know Your Data
Start with repetitive, high volume, taxing tasks
Buy vs. Build
76. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
Business
Stakeholder
domain
knowledge
usesIndustry
Market
Company
Products
Customers
Competition
Price of tea in China
Which way the wind is blowing
The airspeed velocity of an unladen African swallow
77. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
BUSINESS
GOALS
Business
Stakeholder
domain
knowledge
uses
to set
78. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
uses
to set
BUSINESS
GOALS
domain
knowledge
Business
Stakeholder
BUSINESS
PROBLEMS
to identify
79. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
uses
to set
BUSINESS
GOALS
domain
knowledge
Business
Stakeholder
BUSINESS
DECISIONS
to make
to identify
BUSINESS
PROBLEMS
80. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
uses
to set
BUSINESS
GOALS
domain
knowledge
Business
Stakeholder
to identify
BUSINESS
PROBLEMS
BUSINESS
DECISIONS
to make
informed by
Analyst
insight
81. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
uses
to set
BUSINESS
GOALS
domain
knowledge
Business
Stakeholder
to identify
BUSINESS
PROBLEMS
BUSINESS
DECISIONS
to make
informed by
Analyst
insight
Analyst
82. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
uses
to set
BUSINESS
GOALS
domain
knowledge
Business
Stakeholder
to identify
BUSINESS
PROBLEMS
BUSINESS
DECISIONS
to make
informed by
Analyst
insight
Analyst
appreciates and considers
83. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
to craft to test
uses
to set
BUSINESS
GOALS
domain
knowledge
Business
Stakeholder
to identify
BUSINESS
PROBLEMS
BUSINESS
DECISIONS
to make
informed by
Analyst
insight
appreciates and considers
Analyst
HYPOTHESES
86. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
determines
accounts for
most revealing data
cognitive bias
Analyst
innate bias
to
ensure
SOUND
ADVICE
Business
Stakeholder
87. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
determines
accounts for
most revealing data
cognitive bias
Analyst
innate bias
to
ensure
uses
to convey
SOUND
ADVICE
visualization
and storytelling
Business
Stakeholder
88. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
builds
chooses
Analyst
depends on
Data
Scientist
METHODS
MODELS
Data
Engineer
depends on
collects
cleans
integrates
monitors
manages
pipelines
operationalize
s DATA
QUALITY
to improve
give
feedback
Data
Business
StakeholderBusiness
Stakeholder
89. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
builds
chooses
Analyst
depends on
Data
Scientist
METHODS
MODELS
Data
Engineer
depends on
collects
cleans
integrates
monitors
manages
pipelines
operationalize
s DATA
QUALITY
to improve
give
feedback
Data
to ensure quality of
Business
Stakeholder
91. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
Data
Engineer
Data
Scientist
Analyst
Business
Stakeholder
Licorne
92. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
give
feedback
Data
Engineer
Data
Scientist
Analyst
Outsource
if you can
Business
Stakeholder
93. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
AI Onboarding Tips
Clearly identified goals
Know Your Data
Start with repetitive, high volume, taxing tasks
Buy vs. Build
Buy!
94. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
AI Onboarding Tips
Clearly identified goals
Know Your Data
Start with repetitive, high volume, taxing tasks
Buy vs. Build
Determine which data sets are useful
95. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
AI Onboarding Tips
Clearly identified goals
Know Your Data
Start with repetitive, high volume, taxing tasks
Buy vs. Build
Determine which data sets are useful
Too much = noise
Too little = overfitting
Just right = insight
96. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
AI Onboarding Tips
Clearly identified goals
Know Your Data
Start with repetitive, high volume, taxing tasks
Buy vs. Build
Determine which data sets are useful
Become proficient at the Smell Test
98. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
AI Onboarding Tips
Clearly identified goals
Know Your Data
Start with repetitive, high volume, taxing tasks
Buy vs. Build
Determine which data sets are useful
Become proficient at the Smell Test
Be Human
99. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
Amount of data
Speed of correlation
Repetition minus ennui
Accuracy
Cost
Zero attitude
Machine Advantages Your Advantages
Reason
Common sense
Emotion
Empathy
Experience
Integrated cognition
Be Human
100. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
Press Your Advantage
Recognizing the problem
Onboarding new ideas
Relating non-related data
Relating non-related experience
Collaboration & Diversity
Empathy
Imagination
102. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
AI Onboarding Tips
Clearly identified goals
Know Your Data
Start with repetitive, high volume, taxing tasks
Buy vs. Build
Determine which data sets are useful
Become proficient at the Smell Test
Be Human
106. jsterne@targeting.com – Marketing Evolution Experience.com – @jimsterne
Harness the Power for Yourself
Take advantage of the tools
Build systems to talk to customers' systems
Build your brand