5. At its peak in 2006, Second Life had
a story on the cover of Business
Week, a 12-page spread in Wired,
and numerous blog posts about
brands like Coca-Cola, Scion and
even the NBA establishing in-
world presences.
Reuters, CNET, CNN and Wired also
set up virtual news bureaus, though
all …
but CNN killed off their in-
world coverage about a year
later.
10. Mapping
Technologies
Through the
Gartner
Curve
Gartner Hype Cycles are a graphic
representation of "the maturity and
adoption of technologies and
applications, and how they are
potentially relevant to solving real
business problems and exploiting new
opportunities"
31. Interacting With
Human
Acting as Human
Speech
Vision and Hearing
Natural Language
Processing
“Machine” Learning
- Recommendation
32.
33. The 5 “P” of A.I.
Helping to …
“Intelligent”
vs
Relevant
Planning: Building relevant strategies.
Production: Creating relevant content.
Personalization: Powering relevant consumer
experiences.
Promo2on: Managing relevant cross-channel and cross-
device promo;ons.
Performance: Turning data into applicable insights
39. A.I. Is moving 4P’s to S.A.V.EEE.
in a Hyper Personalized Relation
Product Place
Price Promotion
Service x Product
= Bespoke Offer
Value Based
Pricing
Access x
(Learn + Apply)
= Experience
Engage, Entertain,
Educate = Unique
Relationship
40. A.I. Is moving Product to Service
in a Bespoke Offer
Product
Service x Product
= Bespoke Offer
1. Recommendation
2. New Features
3. Churn Prediction
44. A recommenda5on
system generates a
compiled list of items in
which a user might be
interested, in the
reciprocity of their
current selec5on of
item(s).
It expands users’
sugges5ons without any
disturbance or
monotony, and it does
not recommend items
that the user already
knows.
USE CASE 1
50. Without seeing a single episode of House of Cards, NeBlix
commiDed to two seasons of the show, or 26 episodes, bidding a
reported $100 million.
That’s $3.8 million per episode.
Give people what they want, when they want it, in the form they want it
in, at a reasonable price — and they'll more likely pay for it rather than
steal it."
51. By analyzing viewer
data – think 30
million “plays”, 4
million ratings, 3
million searches -
the company was
able to determine
that fans of the
original House of
Cards, which aired in
the UK, were also
watching movies
that starred Kevin
Spacey and were
directed by David
Fincher, who’s one
of the show’s
executive
producers.
52. Churn represents the percentage of lost customers in a given time range
Customer churn is a major threat to every streaming company.
Companies lose $1.6 trillion per year to churn1
USE CASE 3
53. Acquiring a new
customer is anywhere
from five to 25 8mes
more expensive than
retaining an exis8ng one.
Increasing customer
reten8on rates by 5%
increases profits by 25%
to 95%.
54. Reach out in the
moment of max
impact to keep
customers hooked
and wanting more
/ Automate
resolutions to
customer service
issues / Engage in
an ongoing
conversation with
the customers
By analyzing churning customers, companies like
Netflix can start to pick up on profile attributes,
changes in viewing behavior, lifecycle, and customer
service contacts that are likely to lead to churn.
Companies can then preemp5vely intervene to try
and retain an at-risk customer who is demonstra5ng
similar behavior.
For instance, it could be a customer who has reached
out two times in a certain time frame with streaming
or downloading issues. Or, a person who has
abandoned all content that they have started in the
past four weeks without completing any title.
55. $6 billion on content this year
◦ If it just presented the most popular selections to everyone, many titles would remain unseen.
◦ Wasted content spend, Less time watching Netflix instead of other networks.
Take-rate on top personalized video recommendations was 3 to 4 times higher than the most popular
videos.
◦ Viewers were exposed to 4 times as many videos using personalized recommendations compared to a list of
the most popular titles.
Recommendations = surface niche titles that wouldn't find an audience on a traditional cable
network, but that its viewers love
◦ Save money on its content & maximize the value of inexpensive titles.
Recommendations lead to meaningful increases in overall engagement with the product (e.g.,
streaming hours) and lower subscription cancellations rates.
◦ Higher lifetime customer values and lower revenue volatility.
◦ Retention rates are already high enough that it takes a very meaningful improvement to make a retention
difference of even 0.1%.
56.
57. Curated shopping is
one of the firm’s
most important
projects for the next
12 to 18 months, says
Zalando CEO Rubin
Ri_er.
Hyper
Personalized
Relation
64. A.I. Is moving Price to Value
Price
Value Based
Pricing
1. Dynamic Pricing
2. Yield Segmentation
3. Predictive &
Prescriptive buying
Paterns
65. AI and machine learning are
helping pricing managers
capture more revenue and
profits by finding how what a
given customer is willing to
pay or op@mizing price across
their customer and product
mix.
66. Uber: Till 2017 - Prices go up to encourage more drivers to go
online. The increase in price is proportionate to demand
68. Surge Pricing Acceptation Rate
“One of the strongest predictors of whether or not you are going to
be sensitive to surge is how much battery you have left on your cell
phone,” Keith Chen, Uber’s head of economic research said.
In other words, you’re more likely to accept a surge-priced fare, regardless of the price, if your phone’s
about to die because you need a ride home immediately, and if your battery’s death is imminent, you
can’t afford to wait 15 minutes to see if the price drops down again.
Uber knows when its users’ phone batteries are running low because the app switches into power-saving
mode.
“And we absolutely don’t use that to kind of like push you a higher surge price, but it’s an interesting kind
of a psychological fact of human behavior,” Chen said.
69. Using large quantities of data,
AI can also pull from different
sources to add to the
algorithm such as events,
time of day, amount of
people requesting rides and
more.
With pulling in these real-
time pricing dynamics, this
means that pricing can vary
from minute to minute. It can
also pull in personal
customer data and historical
activity to determine how to
price the ride.
If your online behavior shows
a pattern of going certain
places and times, it may
charge more to your account
at those times. Engineering Uncertainty Esfmafon in Neural Networks
(uber.com)
70. Accurate time series forecasting during high variance segments (e.g.,
holidays and sporting events) is critical for anomaly detection, resource
allocation, budget planning, and other related tasks necessary to
facilitate optimal Uber user experiences at scale
71. In General
OpPmal pricing
is calculated
taking into
account several
variables
Competitors’ prices: Competitors’ pricing is a key variable
à Amazon constantly monitors the prices of online
competitors.
Consumer behavior: Tracking the online behavior of
consumers reveals information about their intention to
buy. If a customer keeps returning to a product page, it is
more than likely they are interested in buying. à AI allow
to work individually on an unlimited number of variables.
Shopping periods: Prices can be adjusted according to the
time period when demand for a certain item usually
surges, such as before and during a holiday season such as
Christmas..
72. Dynamic pricing: Retailers using artificial
intelligence to predict top price you'll pay
You could be paying more than others
Data used by pricing algorithms is gathered from seemingly benign sources such as loyalty cards and
even post codes. But it also increasingly incorporates prior behavior.
As with gambling, the house always wins
In the long run, consumers are likely to be the biggest losers, because dynamic pricing plays on our
desire to find that next bargain.
Moral and ethical questions
It has the potential to worsen social inequality, particularly as the determinations used by retail
algorithms are intentionally non-transparent
Algorithms to negotiate on your behalf
Products like Amazon's Echo or Google's Home are the first tangible versions of the quest to make
artificial intelligence butlers a common household feature. The goal is to persuade consumers that
they can save time and energy by delegating mundane tasks like ordering groceries or booking tickets
to these artificial agents.
73. Using AI to identify
then eliminate the
most unproductive
customer discounts
and segments
freeing up more financial
resources and time for those
that contribute to profits.
74. Using AI to iden-fy
then eliminate the
most unproduc-ve
customer discounts
and segments
FREEING UP MORE FINANCIAL
RESOURCES AND TIME FOR THOSE
THAT CONTRIBUTE TO PROFITS.
75. AI is making it possible to create propensity models by persona, and they are invaluable for
predicting which customers will act on a bundling or pricing offer
77. Served grill video and purchased grill.
Served geotargeted patio
message Northern Market.
Served geotargeted patio
message Southern Market.
Served STRONG
promotional message
Went to site
but did not purchase.
Served STRONG
promotional message
Ongoing personalized
messaging
CUSTOMER JOURNEY ANALYTICS DRIVE CONTENT CREATION
78.
79. AI is improving
Configure, Price,
Quote (CPQ)
effectiveness by
bringing greater
accuracy and control
to price
management and
price optimization,
which increases
margins, reduces
costs, and increases
profitable financial
performance.
80. FutureMargin, for
example, is a Shopify App
that uses AI to help
businesses optimize
prices, profit, and
inventory by a variety of
factors, including
predicting demand and
seasonal variations in
products.
81.
82. A.I. Is moving Place to Access to generate
a unique experience.
Place
Access x
(Learn + Apply)
= Experience
1. Last Mile
2. Ux Opfmisafon
3. Usage Suggesfon
4. Machine / Human Collab
83. Useful
Usable
Used
The bot will let you know the current location of your driver and show you a picture of the license plate and
car model.
84. Chatbot for business penetration
is exponential
2016
Regular
Usage
2016
Pilot
2019
Plan
2020
Customer
Service
5% 20% 32% 85%
Source: Gartner / Forrester
85.
86.
87.
88.
89. Kia had more
than 800
websites
where a
customer can
ask about
buying a car
90.
91. What for ?
Better User
Experience
Improved
Customer Service
Data-driven
Product
Recommendations
Automated Lead
Generation and
Qualification
Rewarding
Customer Loyalty
Closing the Deals
97. Automated call centers are expected to
become a reality in 2021 as AI is leading the
revolution of chatbots.
The reason being, AI along with national
language processing will make it possible for
chatbots to be “intelligent” enough to assist
customers like human beings.
103. ASOS Visual Search
Monetize the ‘discovery’
phase of the shopping
journey
Turns the user’s
smartphone camera into a
discovery tool, allowing
them to take a picture of a
product. By identifying the
shape, color, and pattern of
the object, ASOS’ AI
technology can then cross-
reference its own inventory
of products and serve up the
most relevant results.
105. Sephora Visual Artist product à
“try on” cosmetic products
Powered by Modiface AI
technology, the Visual Artist can
map and identify facial
features, then use augmented
reality to “apply” the user’s
selected product and shade.
Moreover, it can automatically
apply suggested shades based
on the consumer’s skin tone.
106.
107.
108.
109.
110.
111. KFC – Voice
Marketing Insights
The brand aims to improve its data
analytics processes by using the
information captured with this
voice-marketing method to learn
more about what customers want.
For example, an uptick on people
ordering a particular kind of
combo meal might trigger KFC
executives to set a discounted
price for it as a limited-time offer.
112. A.I. Is moving Promotion to Engagement,
Entertainment, Education
Promotion
Engage, Entertain,
Educate = Unique
Relationship
1. Content Production
2. Dynamic Creation
3. Insights Automation
4. Micro-Targeting
5. Channel Selection
113.
114. RankBrain is an artificial
intelligence system that
is designed to
understand how people
search and why they’re
searching (or their
intent when searching).
It does this through a
mathematical process
and advanced (and
evolving) understanding
of written language
semantics that’s
intended to help
RankBrain ultimately
“understand” searchers
on a deeper level than
simply matching
keywords
115. 20 % of all business-
oriented content is
generated by machines
(Gartner)
Client reports
Product descriptions
Campaign summaries
TV programming
Announcements Ratings
analysis,
Online content for clients
Social media updates
…
Wordsmith, a software platform lets you
create sophisticated templates that turn
raw data into unique pieces of writing
about that data. Wordsmith is a natural
language generation (NLG) platform.
116.
117. Did a Human or
a Computer
Write This?
MARCH 7, 2015
118. Did a Human or
a Computer
Write This?
MARCH 7, 2015
119. Did a Human or
a Computer
Write This?
MARCH 7, 2015
120. Did a Human or
a Computer
Write This?
MARCH 7, 2015
121. How Virgin Holidays is
using AI to improve
email marketing ROI
In a fairly short space of time,
Virgin Holiday’s AI-optimised
subject lines were generating
more engagement than before.
“We started to see”, Saul says,
“that – for exactly the same
campaigns and exactly the same
segmentation – our open rates
increased two percentage points.”
122. A.I. Men – The
Young Pope:
The Papal Artificial
Intelligence
124. You need to intelligently automate some of the labor-intensive
tasks you do as a marketer.
125. Case 1: Zero Marketing Touchpoints
Let’s start with the easiest scenario: there are zero marketing touchpoints, and you
have sales of $1,000. If you have zero marketing touchpoints, then all $1,000 of your
sales are baseline sales— no sales are attributable to marketing, and attribution is
trivial.
126. Case 2: One Marketing Touchpoint
You want more sales, so you start marketing using touchpoint A and your sales increase to $4,900.
If you used the popular but obsolete and flawed “last touchpoint” methodology, you will attribute all of
the sales to touchpoint A with a value of $4,900. But, this is inconsistent with economic theory, which
says that the value of a good or service is its marginal utility. You already had $1,000 in sales, so
touchpoint A is worth $4,900 – $1,000 = $3,900.
129. The solution is to build an artificial intelligence (AI) that predicts the
outcomes of each scenario based upon sales and touchpoint combinations
observed in historical data.
130. Let’s Meet Lucy
an invaluable
member of the
marketing team
Lucy can
à hunt down facts about a
potential market
à pull together customer profiles
from social media
à draw up media models in
moments.
She performs previously time-
consuming tasks in an instant,
empowering marketers to do
much more in far less time, and
the more she does the more she
learns, growing and developing
according to a campaign’s needs.
131.
132. Harley Davidson use Albert
Autonomous Media Buying A.I.
After processing high-level KPIs, parameters, and creative materials, Albert quickly identified
emerging user behavior and patterns across paid search and social media.
He then analyzed the behaviors to produce insights and suggestions for optimal performance and
scalability. Most crucially, he then autonomously executed on the suggestions, delivering previously
unseen results across channels including:
◦ Proactive creative and messaging insights & recommendations.
◦ Micro-campaigns of infinite campaign variables.
◦ Autonomous media buying using predictive analytics and deep learning.
133. A.I. Capacities = (Human Capacities)
x
A.I. (Albert) can create user
segments based on:
Contacts: People in the brand’s existing CRM,
which can be broken down into segments such
as ‘parents', 'highest lifetime value',' newsletter
subscribers' or 'loyalty club members'.
Visitors: Website visitors who, for example,
added something to their cart, viewed a
product, or made a purchase (these were not
applicable in Dole's case).
Lookalikes: New audiences resembling
previous high-value customers.
Based on its incessant testing, A.I. (Albert) can
determine :
What media to invest in, at what times and in what formats.
How much budget to spend on each channel and for each ad format.
Which creative + headline combinations to use for each micro-segment.
Which ad combinations to place where.
Which cities, audience types and audience behaviours to prioritize and
target.
Which cities, audience types, and behaviours to eliminate.
139. A.I. is already integrated in
Marketing
Consumer is not
completely perceiving that
reality
à We (Human) + A.I. = 3
à Let’s Human
Concentrate on Strategy
and Creative Big Idea
à Test now is a must – Full
productivity is about 10
Years