AI in Marketing

Prof. Zillur Rahman

Department of Management Studies
Indian Institute of Technology, Roorkee
Week-7
Lec 31-AI and New Brand Realities-II

Welcome to this NPTEL online certification course on artificial intelligence and

marketing. And now we talk about module 31 and we are still continuing with AI and
new brand realities that we started in module 30. So, this is the part 2 of that and let us
look at what are the things that we will cover in this module. So, we will start with
discussing the emerging phenomena of brand platforms. Then we will go about
differentiating brand platforms from brand aggregator platforms and then discuss the
mechanism of flagship brand platform formation. And then we will discuss the building
block of brand platforms and related outcomes.

So, we will start with platformization of brands. Digital platforms that aggregate products
and services such as Google Shopping or Amazon has emerged as a powerful
intermediary to brand offerings challenging traditional product brands that have largely
lost direct access to the consumers. As a countermeasure, several long-established brands
have built their own flagship platforms to resume the control and foster consumer loyalty.
For example, sports brands such as Nike, Adidas, SES launch tracking and training
platforms that allow for ongoing versatile interaction among participants beyond product
purchases.

So, the problem with these these platforms is that they take away the customers from the
brands. Brand platforms is an emerging domain in branding. Although the concept of
brand communities and brand aggregator platforms have been relevant for quite some
time. Many brands used to offer allied services as well, but the advent of smart
technology and AI are now allowing brands to form their own platforms. So, now brands
will not have to be dependent on other platforms.

We have briefly hinted at this phenomenon of platformization in the module on network

effects, which is itself an important factor for brand platforms. AI and related smart
technologies are the backbones of these platformization efforts. Let us delve into this new
brand reality and explore what are these brands' platforms and how brands can go about
building their own platforms. So, we will start with understanding brand communities.
Companies are interested in collaborating with consumers through the internet or
otherwise to create value through communities built around brands.

A brand community is a specialized community of consumers and employees whose

identification and activity focus on the brand. So, this brand, this community includes the
consumers and employees. There are three characteristics of brand communities. The first
is a consciousness of kind or a sense of felt connection to the brand, company, product or
other community members. So, there is a connection between the consumers, the
company, the product, the brand and other community members.

The next is shared rituals, stories and traditions that help to convey the meaning of the
community and also they unite the members. A shared moral responsibility or duty to
both the community as a whole and individual community member. Now, what are
digital platforms? Digitization promised established product brands nothing short of their
emancipation from the traditional retailing value chain. Online channels made cutting out
intermediaries and establishing direct consumer access and relationships. Both simple
and inexpensive and many brands willingly embraced this opportunity.

However, when new digital aggregators of products and services emerged, in particular
online marketplaces and the search engines, they rapidly observed the interface to
consumers forcing many offline born products brand back to second rank. What are these
brand aggregation platforms? For example, Amazon, Flipkart, Google Shopping and
[Link] have become important access point for products from household tools to sports
equipments which we use as a recurring example. Leaving brands such as Adidas, Nike
or Asus to resume their basis as suppliers to this new group of digital intermediaries. So,
now these companies are just reduced to suppliers.

These intermediaries which we call brand aggregation platforms differ from conventional
retailers by relying on a platform business model, by relying on a platform business
model where they provide the infrastructure and governance to enable commercial
transactions between external suppliers and consumers of branded products while not
offering these products themselves. So, they provide a platform for sellers and buyers to
come together. For example, searching for Adidas running shoes on Google brings up
Google Shopping pages from numerous retailers as well as Adidas own shop. Google
aggregates these offerings while the consumers transact directly with the brand or an
online retailer. As a result, these platforms provide value by granting user access to a
wide variety of products and services enabling them to organize their consumption
around few powerful interfaces.

This reintermediation through brand aggregation platforms arguably leaves many brand
worse off than in the pre-digital retailing era because these platforms diminish brand
differentiation and foster price competition by featuring many similar or even identical
offerings at different prices from competing suppliers. So, the difference between brands
is getting diminished. For example, search for a sports jersey on Zalando, a former online
apparel retailer turned brand aggregation platform, provides an exhaustive overview of
products from brand-owned shops and myriad retailers. Now, let us look at the
platformization of brands. As a countermeasure to this development, some established
product brands have started to venture into the platform business themselves.

So, now these brands are making their own platforms. So, as a countermeasure to this
development, some established product brands have started to venture into the platform
business themselves either by extended their operation organically that is Nike Run Club
or by acquiring existing platforms Adidas Runtastic, Askis Run Keeper. The
platformization of brands create offerings that transcend the specific product brand by
including third-party complementary products, services and contents to occupy the
broader category space and address consumer needs more holistically. However, brands
have a hard time building competitive flagship platforms. While retailers have long
embraced platformization, for example, Amazon and Zalando Douglas as a natural
evolution of their consumer focused aggregator approach.

Product brands lack cross-product expertise, treat the platform as yet another sales
channel or fear the inclusion of competitors offering to name just a few impediments.
Thus, the platformization of brand is still in its infancy. Though some encouraging
examples have emerged in such markets as Athletics, Nike, Gramen, Do It Yourself, for
example, Bosch, Mobility, BMW and Daimler, Nutrition, Maggie or Gaming, Epic. For
example, Bosch, DIY and Gartner allows consumers to register their tools, engage with
the versatile community and receive advice for DIY projects that is Do It Yourself
projects. In Mobility, BMW and Daimler have extended their share now platforms to
include ride hailing, renting scooters, cars and bikes and finding parkings and charging
spots.

And in Nutrition, Maggie offers online twin classes and events, inspirational contents,
recipes that link directly to supermarkets. In other industries, including business to
business industries, brands gradually acknowledge the potential of building their own
digital platforms. But many have difficulty getting started. So, now this this platform is
not about selling a single or couple of brands. These are about providing a full
experience throughout the journey.

What is the definition of the platform? A platform provides the infrastructure and
governance to facilitate interactions between autonomous agents, interaction between
autonomous independent agents. That is, platforms link agents onto or more market sides
that determine the conditions of their interactions directly. That is, what to exchange, how
to exchange it, maintaining residual control rights over their assets. If Nike offers the
brands merchandise to consumers or Amazon, the seller can typically set prices, control
the item and quantities offered and decide promotions and delivery. Platforms therefore
assume a mediator role, aiming to provide optimal matches between agents on both
market sides, often through digital technologies.

Platform interaction can range from the commercials selling off products and services to
engaging in forums or social communities to posting and consuming media content on
video platforms. Accordingly, examples are abundant and these include social media, for
example, Instagram and TikTok, knowledge exchange forums, for example, Stack,
Overflow, Quora, communities like Reddit, brand communities, video sharing platforms,
YouTube, Vimeo, advertising platform, Google AdWords, service platforms, Uber and
Airbnb, hardware and software platforms, for example, games console and software-
software platforms, operating systems and app stores. Now, let us look at the brand
flagship platforms as a counter to brand aggregation platforms. Brand aggregation
platform has the interaction between market sides is at its core a commercial exchange
between buyers and sellers of branded products. Brand aggregation platforms thus serve
as intermediaries to discrete transactions that lower search cost and efficiently match
customers with products or service offerings.

The transactions are discrete in that purchases do not trigger an ongoing feedback
process other than, for example, dialogues on social media and customer typically return
to enter a new purchase cycle. Despite the widespread success of brand aggregation
platforms, their focus on facilitating transactions across a broad product range also makes
them vulnerable to competitive attacks. They typically lack the resource, expertise,
reputation and infrastructure to individualize customer experience with respect to any
single category and thus fail to occupy a specific category space. By contrast, product
brands can approach value creation through platformization from a more specialized
angle. For example, Nike's Run Club and Training Club platforms are built around
athletics.

Critically, brand flagship platforms as mediators of versatile interaction between

participants are destined to become much more than an own-brand sales channels. So,
these are much more than your sales channel. They offer vast opportunities to co-create
value within the brand's category space through a plethora of commercial and non-
commercial activities. Vertebral interactions may include anything from buying products
to providing and consuming contents. For example, product reviews, creative videos and
educative podcasts or services.

For example, participating in brand or community organized events. Owning to their

versatility, brand flagship platforms may pursue numerous goals such as 1. Increasing
brand awareness and loyalty, 2. Offering complementary products and services and 3.
Triggering consumer feedback processes.

Although the platform may include direct selling of brand inventory, this need not be its
sole or central purpose. RunTastic started out as a tracking application and despite its
acquisition by Adidas in 2015 has preserved and enhanced these features which remain
at the core of the platform's value proposition. Now let us look at the properties of brand
aggregation and brand flagship platforms. So the first column shows the property, the
second is about the brand aggregation platform and the third column is about brand
flagship platform. So the first thing that we that is under properties definition.

Platforms, platform brand owned digital platform that mediates discrete transactions
between buyers and sellers and commercial sellers of branded products. While under
brand flagship platform, product brand owned digital platform that mediates versatile
interactions between participants within the brand related category space. And the key
goals and activities for a brand aggregation platform is that it facilitates direct selling of
products and services. While for a brand flagship platform, there are multiple goals and
activities. For example, creating brand awareness, strengthening consumer relationships,
building community, fostering consumer learning, selling products and services.

The third thing under property is the commercial target function. So, for brand
aggregation platform that means facilitating transactions largely agnostic to product
branding. And for brand flagship platforms, it means directly or indirectly driving brand
sales and fostering brand loyalty. The fourth property is activity scope. Here it is in brand
aggregation platform, it is narrow.

That is product focused aggregators of commercial offerings with ancillary services.

While in brand flagship platforms, it is broad, consumer focused, versatile and multi-
sided network of products, services and contents. The sixth property is the ownership. For
brand aggregation platform, it is online bond platform brand and platformized retailer.
While for brand flagship platform, it is offline bond product brands.

The seventh property is the role of platform owner. For brand aggregation platform, it is
the mediator of commercial transactions. And for brand flagship platform, it is mediator
and supplier of products, services and content. The eighth property is roles of platform
participants. So, in brand aggregation platforms, they are largely well-defined buyers and
sellers.

While in brand flagship platform, they are loosely defined. That is participants can
assume many different roles. The ninth property is the role of product brands. So here it
is, competes for limited platform space and address similar consumption needs as
competing offerings. While for brand flagship platform, it orchestrates value creating
activities on the platform.

The tenth property is competition. In this, it encourages within the brand and between
brand competition. Here it aims to avoid within brand and between brand competition
and focus on complements. The eleventh property is inventory. That is in brand
aggregation platform, internal, none or comparatively limited. While external, it is
extensive assortments of competing third party offerings.

While brand flagship platform, the internal offers own branded products and possibly
external offers complements, possibly direct competitor products. The twelfth property is
user experience. In brand aggregation platform, it means standardized. That is category,
agnostic aims for simple product comparison. While in brand flagship platform, it is
individualized category specific, aims for optimized consumer category experience.

Examples in the first category include Amazon Marketplace, Google Shopping, Wish,
IDLO, [Link], Alibaba and Zalando. In the second category that include Nike's
Training Club, Adidas Runtastic, Askies Runkeeper and Bosch DIY and Garden. Now
how to go about assembling brand flagship platforms? Assembling brand flagship
platform is not trivial as they require strategic and operational choices that may
substantially enhance or limit their value to consumers. In both practice and theory,
brand platforms are still evolving and guidance on how to approach the transformation is
scarce.

A new framework to systemize the assemblage and management of brand flagship

platforms have been gaining transaction and this is decision process outcome framework.
This framework is rooted in the idea for platforms as communities that leverage the
wisdom and addressability of the crowd that is an undefined and generally large network
of actors. Brands can use this framework as a conceptual basis to map the possible paths
to build their own flagship platforms. So here it is about brand decision, consumer
processes and relationship outcomes. So brand decisions are followed by consumer
processes and these are followed by relationship outcomes.

So and then there goes the feedback loop that is based on brand goal alignment. So here
when we are talking of relationship outcomes on the x axis we have consumer crowd
sending and on the y axis we have consumer crowd sourcing and these are the four
different options. That is the level of alignment between desired brand goals and achieved
consumers platform relationship states. These are the various assembling platforms,
building blocks, crowd sourcing from and crowd sending to the platform. This is the
consumer, this is the platform, sourcing from, sending.
Digital platforms as places of crowd sourcing and crowd sending. Digital platform
address consumer goals by relegating them to the open community or the crowd.
Underlying the platform concept is the idea that standing in a crowd and shouting out
why you are there is often more beneficial than reaching out to agents in this crowds in
this crowd individually. The crowd's assets and capabilities collectively exceed those of
any subset of agents. Making scarce resources more abundant and increasing the number
of unique offerings.

Furthermore the search for solution in crowds become less resource intensive because
one market side that is the consumer articulates its goals to all participants on the other
side that are suppliers and that too simultaneously. Consumer crowd sourcing that is
consumers open assignment of a task to a network of people or other entities. On
platforms task assignments pertains not only to consumers but to all platform participants
including third party businesses and product brands which assume similar positions as
contributors to the whole. Consumers harness the power of this crowd rather than
individual retailers or brands whose products services and content offerings are bundled
and made digestible through technology. Notably it is irrelevant if the consumer decides
to source from one platform participant only, follows one training plan posted by a single
person as this choice is the outcome of the crowd sourcing activity akin to the winners of
a crowd sourcing competition.

The consumer selects the offering that suits them best out of all crowd supplied offerings
like strolling through the weekly market to fill their bags. On digital platforms recipients
can also add to crowd solutions themselves. For example, rating products engaging in
discussions and uploading pictures videos and music playlist. Crowd sending is defined
as consumer activity to contribute to the network through the suppliers of product
services or content. Engaging in crowd sending may for example strengthen consumer
social identity and status, bestow a sense of purpose and belonging and allow them to
gain monetary rewards.

Now we will look at this decision process outcome framework. The first is decision.
Consumer goals and platform building blocks. The crowd that is the collective of
platform participants. So this crowd does not mean a collection of useless people but it is
a collection of platform participants.

It is particularly good at addressing important consumer goals which renders digital

platforms a superior alternative to retailers and direct interactions. The questions thus
arises as to when crowds sourcing and sending activities become particularly rewarding
for consumers. There are five key goals that consumers pursue when sourcing from or
sending to the crowd. So the first is commercial exchange, the second is social exchange,
third is self-improvement, the fourth is epistemic empowerment and the fifth is creative
empowerment. Digital platforms cater to these important goals in the form of five
building blocks which are transaction block, community block, benchmarking block,
guidance block and inspiration block.

Importantly following straight from the consumer's goal perspective, these building
blocks do not map exactly to technical platform features. For instance the sharing of a
training exercise could reflect a social desire that is engaging with the crowd but also
serves as guidance, inspiration or a way of benchmarking for users. Thus the same
technical implementation may reflect several building blocks and thus cater to several
consumer goals. Let us look at each of these platform building blocks in detail.

The first is the transaction blocks. Catering to the goal of commercial exchange with a
transaction block focuses on providing matches between demand and supply to ensure
that consumers find the offering that best suit their needs within the assortment of
products or services. Although facilitating commercial exchange is not unique to digital
platforms, the introduction of virtually endless shelf space coupled with advanced
matching algorithm and rapid scalability on both sides that is on the on the sides of
buyers and sellers through network effects have laid the foundation for platform
providing superior customer value relative to traditional pipeline businesses. Digital
platforms allow for almost frictionless participation, provides higher price and quality
transparency and allows buyers and sellers to find the perfect exchange partners easily
and quickly thereby vastly diminishing the transaction costs. For many commercial
platforms such as brand aggregation platforms, the exchange block is at the core of their
operations with additional infrastructure assuming supporting functions.

For example, Amazon marketplace, eBay and Uber. However, some platforms integrate
an exchange block as part of a more balanced approach to so that the commercial
exchange focus does not dominate other platform benefits and thereby prevent further
user growth. The next is the social exchange that is the community block. Social
exchange relates to the goal of engaging in transactions with other platform participants
such as consumers, third parties or employees. These interactions are often part of an
ongoing and evolving dialogue between participants aimed at advancing an array of
social practices. Consumers may achieve these goals on the platform through a variety of
means such as by sharing experiences and personal beliefs.

Depending on which features are part of the platform assemblage and how well they are
integrated, broadly positioned brand flagship platforms are able to cater to social
exchange goals much better than individual brand communities which are limited by
their commercial and brand focus, a relatively small group of active participants and the
lack of diversity. Self-improvement that is the benchmarking block. Consumers have the
fundamental urge to live up to their potential, the desire of becoming more and more
what one is to become everything that one is capable of becoming. Self-improvement can
be seen in terms of self-respect and accomplishment and acquiring excellence. These
aspects underlie the quantification and competition trends that many digital platforms
addresses, especially in the context of fitness, health, nutrition and sports.

Two points of reference that are most relevant for the consumers in terms of
benchmarking and drawing innovation to self-improve the platform crowd and the
consumer's own past performance. This development is fueled by the widespread use of
connected devices such as mobile phones and wearables that can track and quantify
consumer activities, workouts, sleep quality, heart rate and more. Digital platforms often
combine self-tracking and measurement technology with crowd-based motivational
features and gamified experience such as user rankings, virtual badges and opportunities
to share accomplishments. The next is epistemic empowerment, the guidance block.
Consumers empowerment is defined as the strengthening of consumers ability, rights or
authorities to consume or otherwise fulfill their objective as a marketplace actor.

Digital platforms support epistemic consumer empowerment through information on the

products and services they offer but also do more. Some platforms focus on participants
exchange of information as in the case of Skillshare where thousands of teachers can
upload educational videos or Komoot where participants can record hiking or biking
routes and recommend them to others. Other platforms build a strong guidance block
around their main commercial operations by allowing participants to share instructional
videos, blog posts and structured product feedbacks and evaluation. Guidance blocks are
typically standardized, tailored to a specific purpose, one-directional and aim to provide
clear and concise results compared to the more open and socially oriented community
blocks. Owing to the quantity and diversity of content available through crowd-based
guidance blocks, epistemic empowerment of participants on all market sites easily
supersedes what retailers or single brands can offer.

The next is creative empowerment that is the inspirational block. With creative
empowerment we denote satisfying customer curiosity and longing for new experience.
Whereas epistemic empowerment refers to objective and functional knowledge such as
that for comparing product and alternative for purchases creative empowerment refers to
knowledge and stimulation that inspires the consumers. Consumers can use digital
platforms to gain or provide elements of creative empowerment by consuming or creating
content including entertaining content and products. Platforms such as P interest and
Wattpad leverage the crowd to provide a richer set of ideas for consumers seeking
inspiration and a large and engaged group of potential recipients for consumers sharing
their ideas. Nike's Run Club platform induces consumers to inspire and be inspired by the
crowd through music playlist or workout videos that can add for others to use their own
routine.

So now assembling the building blocks. So these transaction block, community block,
benchmarking block, guidance block and inspiration block and then comes platform
assemblage exemplary assemblage 1. So it is ITC. So these assemblages are various
combination of these TC, B, G, I. So the first one is Etsy and Airbnb.

There the transaction block is the biggest one. In assemblage 2, the guidance block is the
biggest one. So that is Komota and Skillshare. In assemblage 3, this benchmarking block
is the biggest one.

So Adidas Runtastic. So you can benchmark with other runners. And in assemblage 4,
this C that is community block is the biggest one. For example, the P interest and
Wattpad. The five building blocks can be reassembled to represent different types of
platform observable in the market. Thinking about the heterogeneous platform landscape
as different assemblages of these building blocks presents a powerful tool for brands in
their creation of their own flagship platform because it helps them select the optimal
assemblage to achieve specific brand goals. The existence and the relative scope of each
building block jointly determines which platform type emerge.

For example, an emphasis on transaction blocks tend to produce marketplaces that cater
primarily to the goal of commercial exchanges. A strong benchmarking blocks yields
tracking and coaching platforms that cater primarily to the goal of self-improvement.
Other platforms combine blocks to fulfill several consumer goals at the same time and
allow each consumer to pursue their own goals, choosing a specific consumer platform
sub-assemblage. For instance, Ryan Nikes Run Club offers tracking and ranking features
to support self-improvement and sharing features to socially engage with other
participants while the consumers are either syncretic crowdsourcing and sending
activities yielding individual sub-assemblages.

The brand sets the general frame through its assemblage of building blocks. Digital
platforms are revolutionary not only because of their ability to to scale quickly and
leverage network effect but also because of their versatility and integration ability. They
create greater value for consumers by allowing them to draw from and provide to crowd
using assemblages of building blocks that facilitate idiosyncratic goal pursuit. This is
important for brands because it allows them to counter the intermediation through brand
aggregation platforms by evading direct competition on their home turf and instead
establish a direct interface with consumers, developing relationships that go beyond
discrete transactions. The outcome that is consumer platform relationships. The
relationships that emerge from flagship platforms reflect focal outcomes of brand
platformization efforts because they determine how brands can leverage platforms
strategically.

By relationships, we denote the possible states that consumers may enter in their
interactions with brands flagship platforms and that vary in terms of their self-relevance,
commitment and durability. As consumers interact with platform participants through
crowdsourcing and sending activities, the nature of these interactions shape the
relationship state that emerge. This effect develops in two steps. The first is consumer
crowdsourcing and sending intensifies as relationship determinants.

The second is aligning consumer platform relationship with brand goals. The first step
establishes the general link between the consumer processes of crowdsourcing and
sending on the one hand and the consumer platform relationship on the other. The nature
of interactions on brand flagship platforms varies by the intensity of consumer
crowdsourcing and sending and that this intensity increases as consumer use the platform
to pursue higher level goals. This intensity in terms affect consumer platform
relationships on the spectrum from discrete exchange rate transaction to versatile ongoing
interactions. More intense crowdsourcing and crowd sending moves the needle towards
the versatile end of the spectrum that is increasingly profound relationships that
culminate in a deep embeddedness of the platform in consumers life. The second step
separates crowdsourcing and crowd sending intensities to constitute key dimensions of a
typology of archetypal relationship states.

This step helps brand analyze specific outcomes of their platform assemblage efforts. It
directs which platform state brands should foster on their flagship platform to realize
particular brand goals as well as which risks are involved. So, to conclude in this module
we have discussed the emerging phenomena of brand platformization as a counter to
brand aggregate platforms. Then we have discussed the five building blocks of brand
platforms and the assembling of these blocks to give different types of brand platforms.
And finally, we have discussed the decision process outcome framework of brand
flagship platform formation. And these are the six sources from which the material for
this module was taken. Thank you.
 AI in Marketing
Prof. Zillur Rahman
Department of Management Studies
Indian Institute of Technology, Roorkee
Week-7
Lec 32-AI for Value Creation and Product Development

Welcome to this NPTEL online certification course on artificial intelligence in Marketing and
now we will talk about module 32. In this module as you can see from this slide, we are
talking about using artificial intelligence for value creation and product development. So
this is the module overview. We will start with understanding what is value. Then we will
move on to understand the current scenario of AI adoption by the industry. Thereafter, we
will understand how to digitize product development and then how is AI benefiting product
development process.

Now let us start with understanding what is value in marketing. Value is all the perceived
benefits tangible and intangible minus all the perceived cost again tangible and intangible
incurred in obtaining, using and disposing of the product or service or any offering. So it is
perceived benefit all tangible and intangible minus perceived cost again tangible and
intangible while you are obtaining, using and disposing of an offering. And this is customer
perceived value that is CPV.

As you can see that this is the total customer cost and this is the total customer benefit. This
benefit is made up of, of product benefit, service benefits, personnel benefits and image
benefits. While the costs are made up of monetary cost and these three are non-monetary
cost, time cost, energy cost and psychological cost. So total customer benefit, the perceived
monetary value of the bundle of economic, functional and psychological benefits customer
expects from a given market offering because of the product, service, people and image.
Total customer cost is the perceived bundle of cost customer expects and incur.

Customer expects to incur in evaluating, obtaining, using and disposing of a given market
offering including monetary, time, energy and psychological costs. Customers perceived
value is thus based on the difference between benefits the customer gets and the cost he or
she assumes for different choices. The marketers can increase the value of the customer's
offering by raising, increasing economic, functional or emotional benefits and or reducing
one or more of the costs. Now we will look at the adoption of AI for value creation by
industry. In 2019, start-ups raised $16.5 billion for AI-based new product development.
One in ten firms uses ten or more AI applications like chatbots, process optimization, etc.
The size of the connected products market will be between $519 billion to $685 billion due
to AI and ML-enabled services. Consolidation of Internet of Things platform market due to
rapid advances in AI-based apps, products and services. The advantage is 14% of
enterprises who are the most advanced using AI and ML for new product development earn
more than 30% of their revenues from fully digital products or services and lead their
peers in successfully using nine key technologies and tools.

PwC found that digital champions are significantly ahead in generating revenues from new
products and services and more than a fifth of champions, 29%, earn more than 30% of
revenues from new products within two years of information. Digital channels have high
expectations for gaining greater benefits from personalization as well. The following report
from PwC compares digital champions' success with AI and ML-based new product
development tools versus their peers. So as you can see on the right hand side, so this is the
overall sample and this is the digital champions. So with co-creation, the gains are 76%
process simulation, 74% product portfolio management, social listening 67% and product
life cycle management is 52% and these are in the last three years and these are today.

Similarly, on the right hand side where we have digital champions, so this light pink is in
three years and the dark are today. Again, you can see that for value co-creation for digital
champions it is 4% and 71%. For process simulation it is 82% and 14% and for digital
twins it is 61% and 19% and product life cycle management it is 61% and 10%. So there
has been an increase, overall increase as you can see from the from this slide there has been
an overall increase also. Most are using at least a few key tools to some extent and the
majority of digital champions are already using most tools.

Product development teams most advanced in AI and ML achieve greater economies of

scale, efficiency and speed across these three core areas of development. Tools fostering co-
creation stand out as the tools currently getting the broadest uptake with nearly two-third
of companies using them already. And around half of the companies already use product
portfolio management, process simulation and digital prototyping. Far fewer companies
are using agile product development methods, digital twins or PML software although
interest in all of these is growing. Digital champions are noticeably ahead in implementing
these technologies with more than 60% using all three.

Product development teams most advanced in AI and ML achieve greater economies of

scales, efficiency and speed across these three core areas of development. And the three
core areas of development is the first is concept and specification, the second is design and
development and the third is test and go to market. Now on this on the x axis we see least
effective tools and the most effective tools. So at the first stage that is concept and
specification the least effective tool is the digital prototyping and then the most effective is
product portfolio management and in between we have product life cycle management and
co-creation. So co-creation is coming from social listening, product life cycle management,
process simulation then there are digital twins.

At the second stage that is design and development the most effective tools are co-creation,
digital twins and agile product life cycle management, data analytics and AI, digital
prototyping. At the third stage the least effective tool is the product portfolio management
that is digital prototyping and the most effective tool is data analytics and AI. Now let us
look at the digital twin and the case of Electrolux. Electrolux is a leading global appliance
company that places the consumers at the heart of everything they do. Through their
brands including Electrolux, AEG and Frigidaire the company sells more than 60 million
households and professional products around the world.

Developing products that meet customer needs in each of Electrolux more than 150
markets is not easy. A decade ago the company met that challenge by using a wide range of
diverse product development tools around the world. Today Electrolux is achieving greater
efficiency in product development through its global implementation of one of the
comprehensive PLM tool across the company. Engineers in all of Electrolux locations can
now easily share information relevant to the development of product from technical
documentations to specifications, article acceptance, engineering changes and more. So
they can drive great consumer experience more quickly.

The technical backbone helps Electrolux to get the most out of the company's
implementation of a modular product architecture. Global modular platforms are helping
the company spread successful launches from one market to another with adaptation to
local preferences. Local engineers in each country can easily access digital twin of
Electrolux product masters and translate these into specific modules which are tailored to
the requirements of each individual market. In addition to mirroring all of the information
about the product's physical characteristics, these tools also include any software that has
been integrated into the product, the results of any lab testing etc. so that there is one
single source of truth about a particular product category.

Suppliers can be connected via the platform too so they so that they receive the latest
information about any engineering changes immediately and are always up to date. The
third digital twin technology is used in the following four stages of the operations. The first
is the design that is product development twin. The second is plan and make, production
and supply chain twin. The third is maintained operational asset twin and the fourth is
manage that is product lifecycle twin.

Now let us look at the first design that is product development twin. It is a replica of the
product during the product development process. It allows digital testing and simulation of
the product's performance to optimize the performance as well as behavior of a product in
advance. In the ideal state, the digital twin in product development will completely
substitute the physical testing of the product. The second step is plan and make that is
production and supply chain twin that allows simulation and tracking of all processes and
therefore optimization of production and supply chain processes during the product
development process as well as during serial production.

In the ideal state, all processes can be simulated in real time or before the actual process is
conducted. What if and scenario modeling can be used to configure or optimize processes
to derive strategic, tactical and operational decisions. The third is maintained operational
asset twin. It is a replica of the entire production and logistics related assets ecosystem as a
whole, facilitates monitoring and managing of maintenance operations and assets in real or
the right time. In the ideal state, location, condition, health and performance can be tracked
and analyzed in the real time.

The fourth is managed at a product lifecycle twin is a real time 100% replica of the product
at any given time includes all key information generated about the product along the entire
product lifecycle. It facilitates to seamlessly retrieve all product related data. In the ideal
state, product behavior can be completely simulated in advance along the product lifecycle.
How to digitalize product development? The step 1 here is to understand and benchmark
digital performance and capabilities. The first step should be gaining a comprehensive
understanding of current performance in product development and engineering.

How our company's performance compared to competitors within each of the key
capabilities and based on product development KPIs. In addition, we need to assess the
current digital product development capabilities including differentiating capabilities and
pain points challenging when it comes to digital performance. Which capabilities are must
have and which capabilities truly differentiate our company from the competitors. For
example, do we have a stronger understanding of our customers, or do we already use AI to
gain better customer insights? Finally, we need to assess the current pain points around.

1. Digital tools 2. Data analytics 3. Partnerships 4. Digital methods and processes 5.

Existing R&D organization The second step is to detail future capabilities, priorities and
business case. Once we understand where we stand with digital capabilities, we need to
detail which digital capabilities we need to build or enhance to ensure sustainable
performance.

Key focus areas should be defined by carefully looking at our differentiating capabilities.
What capabilities makes us different from the competitors? The key focus area should be
focusing on the differentiating capabilities and our performance to our competitors and
overall as our pain points as a whole. Digital capabilities must include data analytics and AI,
agile development methodologies or implementing digital co-creation tools. To prioritize
our focus areas and to ensure that we have clear targets during the implementation phase, it
is essential to develop a detailed business case for each focus area. A robust cost-benefit
analysis will then help us to prioritize the focus areas and to omit trendy technologies with
limited benefits or of strategic value.

The third step is develop capabilities and tools in an agile and integrated way. During the
next phase of the digital journey, the company needs to develop and implement digital
solutions from capabilities to tools and methodologies in an agile way. For that, it has to
start with the focus area that has the highest priority, then develop and implement the
solutions incrementally until it reaches the targets. Set two-wick sprint goals that provide
with continual results and checkpoints so that it can change the development focus to
achieve more benefits or fail fast if the value of the solution cannot be achieved. Low-fi
prototypes that demonstrate the solution to users, scrum sessions and other tools can
support the company in rapid and successful solution development.

A certain degree of flexibility within the target should be allowed as long as solutions
continue to meet the key business requirement. The fourth step is to roll out and train
engineers in digital capabilities and tools. Once we start implementing new capabilities and
incremental solutions, it is also critical to set up a digital communication program that
informs and educates the engineers about the digital changes and new ways of working
managers. We will need to offer digital training courses both online and offline and face to
face to train and employees as data scientists, digital design engineers or digital PLM. And
in-house digital training academy paired with outside training providers enable a smooth
digital transformation of R&D and engineering departments.

A challenging but realistic roll out plans needs to include the entire regional and global R&D
and engineering organizations in a step-by-step approach. The fifth step is implementing
with partners an agile and integrated product development ecosystem. As a final step
towards becoming a digital champion in product development, the company needs to
integrate its product development activities with those of the key development product and
service partners and lead an integrated development ecosystem. So we are looking at this
integrated development ecosystem not isolated development ecosystem. Through co-
creation and joint product offerings with partners, companies will be able to offer
customers fully integrated complete solutions rather than individual products and services.

Effective co-creation tools, common standards and simultaneous access to joint product
development platforms will become part of this integrated digital ecosystem. Digital
transformation is never easy but for most companies it is essential to stay competitive. To
help the transformation run more smoothly, it is recommended to involve the key
stakeholders and influencers throughout the entire process to ensure their buy-ins. All of
these steps will help the company to move from traditional product development to a truly
digital product development drawing on the strengths of digital tools, data analytics, agile
processes and other key methodologies making the company agile, collaborative, AI driven
and customer centric. Now let us look at this transformation traditionally to digital product
development.

Now from long product life cycles to short product life cycles. So, the transformation has
been from long product life cycles to short product life cycles. Product maturity reached
during production ramp up and here the simulation prototyping enhances pre-launch
maturity. W and V waterfall-based releases of entire systems and here it is agile with
continual releases. Then we have customers interaction from 1 is to n.

Here it is individualized, personalized interactions and products. Then we have siloed data
model. Here we have one E to E product data network along the life cycle. Then we have
the sequential development and design. Here we have ecosystem approach, parallel
development and collaboration.

Here we have product and service improvements based on field performance surveys and
warranty service information and here we have continuous product and service
improvement through constant monitoring and prescriptive analytics. Then post launch
cost savings and manufacturing improvement initiatives have both to pre-launch cost
savings and manufacturability improvement. So, this is the digital product development,
and it consists of digital tools, data analytics, partnerships and ecosystems, methods and
processes and the organization. So see how all this things have transformed to this. Then
we come to the product and technology portfolio.

These are the various steps. Pipeline management, product life cycle management,
simulation analytics, digital twins, co-creation with suppliers and customers, open
innovation, E to E integration, agile development, digitize and simultaneous process, user-
centered design, governance and responsibilities, organizational structures and interfaces,
resources and competencies. So, these are the product and technology portfolio. Again, we
have those four, digital product development, partnerships and ecosystems, data analytics,
digital tools, organizations and methods and processes. Now this leads to digital operations
ecosystem, connected execution and integrated and continuous planning. And that is
individualized solution offering, customer solution ecosystem, this is technology and people
and culture.

So multichannel customer interaction, IT infrastructure and architecture, digital

technologies, career development skills and relationship mindset. So, this is how whole of
this digital product development system it comes together. How AI benefits product
development? We will discuss how AI is assisting enhancing various processes in the
development of a product with examples from the industry. With AI being used, it would
add more value in every process and thus to the final product as well, which eventually
would be reflected throughout increased customer satisfaction. So the first is eliminating
the roadblocks to getting new product launched.

It starts with using AI to improve demand forecasting accuracy. Honeywell is using AI to

reduce energy cost and negative price variance by tracking and analyzing price elasticity
and price sensitivity as well. Honeywell is integrating AI and machine learning algorithms
into procurement, strategic sourcing and cost management, getting solid returns across the
new product development. The next is the AI-based techniques to create and fine-tune
propensity models that define product line extensions and add-on products that deliver the
most profitable cross-sell and upsell opportunities by product line, customer segments and
personas. Propensity modeling is a set of approaches to building predictive models to
forecast behavior of a target audience by analyzing their past behavior.
Propensity models are based on imported data built in Microsoft Excel, making their
ongoing use time-consuming. AI is streamlining creation, fine-tuning and revenue
contributions of upsell and cross-sell strategies while automating the entire progress. All
firms of a US-based TV data company used AI and managed to optimize the spends on
advertisements by predicting the conversion of new users and targeting those with a higher
likelihood to subscribe on premium. The accuracy of predictions was increased from 8 to
80%. AI is enabling the next generation of frameworks that reduce time to market while
improving product quality and flexibility in meeting unique customization requirements on
every customer order.

AI is making it possible to synchronize better suppliers, engineering, development

operations, product management, marketing, pricing, sales and service to ensure a higher
probability of a new product succeeding in the market. Leaders in this area include BMC's
Autonomous Digital Enterprise. This BMC's Autonomous Digital Enterprise framework
shows the potential to deliver next-generation business model for growth-oriented
organizations. Looking to run and reinvent their business with artificial intelligence,
machine learning capabilities and deliver value with competitive differentiation enabled by
agility, customer centricity and actionable insights. The third is forecasting demand for new
products including the causal factors that most drives new sales in an area AI is being
applied to today with strong results.

From the pragmatic approach of asking channel partners, indirect and direct sales teams
how many of new product they will sell to using advanced statistical model, there is a wide
variation in how companies forecast demand for a next-generation product. AI and ML are
proving to be valuable to taking into account causal factors that influence demand yet had
not been known of before. Denonni use machine learning to predict demand variability and
planning. The new capability improved its forecasting process and lead to more efficient
planning between different functions such as marketing and sales. This has led to 20%
reduction in forecast error and a 30% reduction in lost sales.

The next is using AI to analyze and provide recommendations on how product usability can
be improved continuously. It is common for development ops, engineering and product
management to run A-B tests and multivariate tests to identify the usability features,
workflows and app and service responses customers prefer. Based on personal experience,
one of the most challenging aspects of new product development is designing an effective,
engaging and intuitive user experience that turns usability into a strength for the product.
When AI techniques are part of the core new product development cycle including usability,
delivering enjoyable customer experience becomes possible instead of a new app, service or
device. As a chore to use, AI can provide insight to make the experience intuitive and even
fun.

Using generative design algorithms that rely on machine learning techniques to factor in
design constraints and provide an optimized product design. In May 2018, the computer-
aided design software maker Autodesk announced the alliance with General Motors to
explore the use of generative design and additive manufacturing. In developing future
vehicles. Dubbed Project Dreamcatcher Autodesk's generative design algorithm utilizes
machine learning techniques to generate thousands of design options based on the
designer's input such as functional requirement, material, manufacturing methods and
other constraints. The outcomes can be optimized for weight or other performance criteria
which often results in complex geometry suitable to be manufactured using AM techniques.

General Motors Dreamcatcher system used machine learning to transform prototypes. The
solution was recently tested with the prototyping of a seat-welt bracket part which resulted
in a single piece design that is 40% lighter and 20% stronger than the original 8 component
design. So this is a 3D printed seat bracket which is 40% lighter and 20% stronger. So
earlier we had this 8 part now it is and this is a generative design 150 plus design 1 part
consolidating 8 component into 1.

So these 8 components are now converted into 1 component. So to conclude in this module
we briefly discussed about how AI is being adopted by the industry for product
development. Then we have discussed a case study of how Electrolux is using the digital
twin technology. Then we have understood the process of de-utilizing product
development and finally we have discussed how AI is helping the product development
process to provide more value. And these are the 7 sources from which material for this
module was taken. Thank you.
 AI in Marketing
Prof. Zillur Rahman
Department of Management Studies
Indian Institute of Technology, Roorkee

Week 7
Lecture-33 Personalization and hyper-personalization Using AI-I

Welcome to this NPTEL online certification course on artificial intelligence in marketing

and now it is time for module 33. Now as you can see from this slide we are talking about
chapter 6 where we are trying to use AI for designing, delivering and communicating
value and now we will talk about personalization and hyper personalization using AI. So,
this is part 1 of that. This is the module overview. So we will start with how AI is helping
in product personalization followed by how data from various sources can be used to
personalize customer experience through the case of Brinks Home. Then we will see the
limitation companies face in providing personalized customer experience followed by the
ways to overcome the limitation and the introduction to intelligent experience engine.

Now let us look at product personalization using AI. Companies across industries are
putting personalization at the center of their enterprise strategies. So, this is becoming
very important, personalization. For example, Home Depot, JP Morgan Chase, Starbucks
and Nike have publicly announced that personalized and seamless channel experience,
omni-channel experience are at the core of their corporate strategy.

So this is Nike by you. So, these are different kinds of shoes that are that are customized
for you. Now we will see the case of Brinks. Brinks is a 163-year-old business well
known for its fleet of armored trucks. The company also licenses its brands to a lesser
known independently operated sister company Brinks Home.

The Dallas based smart home technology business has struggled to gain brand
recognition commensurate with the Brinks name. It competes against better known
systems from ADT, Google Nest and Ring. Although it has earned stellar reviews from
industry analysts and customers, its market share is only 2%. But its systems have
generated a wealth of product usage information, voluminous historical customer level
transaction data and its field reps have been gathering competitive data since its
beginning operations in 1994. Brinks wanted to find a way to use all this information to
accelerate growth and optimize every customer touch point across all channels, especially
in its messaging, personalization and delivery of user experience.

Implementation of AI In the fall of 2020, working with OfferFit, an AI startup, the

company tested thousands of combinations of messages and offers, varying the creative
content, channel and delivery times. It recognized its structures around customer
acquisition, service and renewal and become and began using AI to optimize service call
scheduling, help cross-sell recommendations from call center reps and conduct customer
outreach for wireless system upgrades. The result in less than 2 years, Brinks increased
A-B testing from 2 to 3 test a day to roughly 50,000 with the capability to add more as
needed. This process has dramatically reduced the need to wait for test results and has
allowed Brinks to personalize every customer touch point. During the first half of 2021,
its average direct to consumer package size increased from $4.89 to $9.68. DTC revenues
per user increased from average of $42.24 to $45.95 during the same period.

Overall revenues increased 9.5% compared to the same period in 2020. Brinks Home is
just one example of how brands can win by tapping a deep store of customer information
to transform and personalize user experiences. From the pre-internet dawn to the
customer journey of the digital era, personalized customer experience has un-equivocally
become the basis for competitive advantage. Personalization is now much more than
what we think.

Customerization now goes far beyond getting customer's name right in the advertising
page, having complete data at the ready when someone calls customer service or tailoring
a web landing page with customer relevant offers. It is the design target for every
physical and virtual touch point, and it is increasingly powered by AI. Over the past 5
years, there have been an increase in their revenues of 6% to 10% and increase in net
incremental revenue attributable to personalization initiatives of anywhere from 40% to
100%. A joint survey conducted by Google involving thousands of customers
immediately following a personalized brand experience revealed a comparable revenue
effect. Companies across all industries are putting personalization at the center of their
enterprise strategies.

Companies in home improvement such as Home Depot in banking like JP Morgan chairs,
in the restaurant industries like Starbucks and in apparels like Nike have publicly
announced that personalized and seamless omni-channel experiences are at the core of
their corporate strategy. We are now at a point where competitive advantage will derive
from the ability to capture, analyze and utilize personalized customer data at scale and
from the use of AI to understand, shape, customize and optimize the customer journey.
Digital advantage supremacy have gone well beyond the boundaries of traditional
marketing to become a much more broader C-suite issue. The obvious winner have been
the big tech companies which have embedded these capabilities in their business models.
But we also see challenger brands such as Sweetgreen in restaurant and Stitch Fix in
apparel that have designed transformative customer experiences based on first party data.

What are the impediments to personalization? Most brands don't personalize customer
experience at the scale or depth necessary to compete with the world's leading companies.
Personalizing an end-to-end customer experience requires orchestration across channels,
a capability that no brand has fully mastered yet. But merging the flow of customer's
physical and digital experience may be the only way challenger brand can compete
against digital natives like Amazon and Google. Early movers have tapped into newer
technologies such as the Internet of Things, Machine Learning, marketing tech platforms
and growing number of digital media tools that can create formidable advantage when
combined with agile methods. Brands that want to surpass or simply catch up with early
movers need to think about their data and technology foundations.

Despite the dizzying array of software tools that purport to enhance every aspect of
customer experience, no one platform can comprehensively manage end-to-end
personalization. Nevertheless, key problems such as creating a 360-degree view of the
customer are being solved with automation, AI-powered intelligence and activation tools
for delivering AI-driven recommendations. The telecommunications giant Comcast used
Pointy List, a customer journey analytics service that locks each customer's footsteps
across its ecosystem. The service time stamps visitors' interaction and generates maps of
each journey. Using AI to gather data and determine where journeys are failing, such as
with its mobile app, Comcast quickly tackles experience issues.

Now what is the new era? Businesses are combining multiple AI; market technologies
and back-office solutions connected through common application programming interface
to better develop and use personalized data. Salesforce and Adobe provide channel
delivery solutions. Customer data platforms such as M-Parity and M-Particle help resolve
identity issues, offer optimization engines such as formation and offer fit, help improve
each ensuing offer and platforms for content generation such as Persado for creative copy
and Sky Day Sky for video enable personalization at scale. New digital media creates
new ways for users to interact with brands. Location-based tracking and payment system
activated by a swipe of a hand blur the line across pre-purchase, that is, advertising and
marketing, purchase, sales and transaction and post-purchase, service and loyalty
interactions.

These capabilities have created intelligent ways to reshape customer experiences and they
enable brands to be distinctive, valuable and deepen engagements. Starbucks for
example, Jio targets labs customers who are near its store with ads about new seasonal
beverages and it sends customers personalized in-app offers to encourage them to visit a
store or try their convenient mobile order and pay option. Most companies don't have the
bandwidth, resources or technical prowess to compete with the likes of Comcast or
Starbucks. The best approach for a challenger brand is to develop a data and tech
roadmap with granular requirements tied to specific customer-driven use cases. For
example, company will need to figure out which customer data elements must be used in
real time to power recommendations in the app or it must determine which systems must
talk to each other after a booking is made to suggest relevant add-on services.

Then it must bring together the business and tech teams to work iteratively focusing on
delivering value as they build the foundation. The limitations of this are in the most
successful digital transformations of the past decade. We have observed that we call the
70-20-10 rule. 70% of the efforts for changing an organization, its processes, ways of
working, key performance indicators and incentives involve people. 20% entails getting
the data right.

The remaining 10% above the technology foundation. Through this we would discuss
four limitations which needs more attention. The first is, most companies are still set up
to be product first, not customer first, making it impossible for the dozen or so teams that
cover channel, market and product silos to collaborate. The increasingly popular role of
Chief Customer Officer was created to solve this problem, to orchestrate the people and
moving parts behind the many customers' touchpoints. Second, analytics is not infused
throughout the business and no single platform integrates customer data and enables
advanced analytics.

And the third is, content is created manually and not tagged for reuse. Finally, agile ways
of working, even if common, in IT teams are generally not used by cross-functional
teams. Without tools to facilitate teams' rapid experimentation and learning, companies
end up with inconsistent, stagnant experiences across channels. Now how to overcome
them? To get started, companies must launch self-governing pods of workers from
marketing operations, analytics, technology and commercial functions and invest in them
with clear goals, budgets and decision rights. These integrated groups should be tasked
with developing a limited number of specific experiences that represent breakthrough
opportunities to drive revenue and build deeper customer bonds.

They should have the tools to measure their day-to-day progress and should work in
intensive two-week sprints to develop and test ideas for improving engagement. They
should optimize many variables, such as what triggers to respond to, which channels to
use, when to reach out to a customer, what messages to issue and what incentives to
offer. AI can play a progressively bigger role in this effort as more experiments are run
and more data is gathered. The pods can use machine learning to determine how to set up
multivariate tests, keep track of everything in motion and decide when to lock in and
scale a test to a broader population. How to build an intelligent experience engine? To
fulfill every goal the customer may have for an end-to-end experience, companies must
think through how to design the flow of a given moment, the information needed to
support it and the cross-channel or cross-party connections, for instance, between in-store
and online or in-mid or post experience required to successfully complete the interaction.
This is not just an exercise in journey mapping or technology planning, it is about
developing the front-end flow to the customer and back-end fuel to drive intelligent
experience engines. Intelligent experience engines are not built just at the highest level of
an end-to-end experience such as enabling better security service at Brinks. They must
also be surgically focused on micro-goals, positive individual movements that compose
the entire experience and ensure that all these goals get stretched together. Moreover,
these engines are intelligent in more than one way. They are crafted creatively and
insightfully using the best possible data and expertise.

And they employ ever improving machine learning algorithms to figure out the right next
step to enable the customer's progress, constantly testing, always learning and fueling
decisions about how the interaction works. What the customer gets is a seamless, positive
and distinctive experience that will only improve over time. The brands that have had the
most success pursue five pivotal practices which define the craft of building intelligent
experience engines. The first is they collect data signals and insights from a constantly
expanding range of sources. Then they reimagine the end-to-end experience as a
seamless flow powered by automated decisions.

Then they activate the experience across channels, connecting touchpoints to engage
customers wherever they may be. So, to conclude in this module, we have briefly
explained how AI is helping the personalization for customers. Then we have talked
about, then we have studied the case of Brinks Home in which the company leveraged on
its huge volume of data to compete with industry giants like Google Home etc. We have
discussed the limitations that companies face while providing personalization. And
finally, we have learnt how to overcome the limitations and the introduction of
intelligence experience engine.

To be continued and that will be continued in the next module. These are the six sources
from which the reference for this module was taken. Thank you.
 AI in Marketing
Prof. Zillur Rahman
Department of Management Studies
Indian Institute of Technology, Roorkee

Week 7
Lecture-34 Personalization and hyper-personalization Using AI-II

Welcome to this NPTEL online certification course on Artificial Intelligence and

Marketing. And now we will talk about module 34. So, as you can see from this slide, we
are continuing our discussion on personalization and hyper-personalization using AI.
And this is part 2 of this topic, personalization and hyper personalization using AI. And
these are the things that we will cover in this module. So, we will start with
understanding how to build an intelligent experience engine using 6 pivotal practices that
is continued from the previous module that is module 33.

And then we will move on to understand the changing market scenario that are leading to
hyper personalization. Then we will understand what hyper personalization is and the
CMOs goal and benefit when they are transformed to hyper personalized marketing. And
then we will understand ways in which data can be collected for hyper personalization
campaigns. So, let us start with understanding product personalization using AI.

So building an intelligent experience engine, these are the following are the 6 pivotal
practices that define the craft of building intelligent experience engines. The first is
connect data, signals and insight. The second is to reimagine the end-to-end experience as
a seamless flow. The third is to activate the experience across channels. The fourth is to
fulfill according to the customer's context.

Then test relentlessly and then continue to hone the craft. Now the first building block of
the intelligent experience engine is to connect data, signals and insights. The first
requirement for building an intelligent experience engine is constructing a 360-degree
view of each customer. Using the expanding range of possible ways to capture new
signals from each one. And how can we capture new signals from each customer? The
athletic apparel company Lululemon invested heavily over the past 5 years to achieve this
goal.

When a guest makes a purchase at a retail location for the first time, she is asked to
provide her email address to receive a receipt. Lululemon, like many other brands,
Lululemon uses this personal information to augment basic customer demographics from
a service like Experian or Axiom, enabling marketing actions such as gender and geo-
based targeting. As people continue to engage with the brand, they often download the
app or shop online. In clicks, stream data is used to understand which items a customer
browsed, which ones they spent a long time considering or came back to and which ones
they quickly moved past. So that all this information will show where does the interest of
a particular customer lies.

This data can be leveraged to infer intent and target future recommendations accordingly.
In 2020, when Lululemon acquired Mirror, it gained a new window into customers'
behavior. Mirror streams fitness classes into users' home, giving Lululemon insights into
customers' workout routines, preference data that helps the brand further refine
recommendations for future products and services. So now you see that this is an apparel
company and it acquired Mirror that is fitness classes. So that was one step to get a 360-
degree view of the customer.

The second building block is reimagining the end-to-end experience as a seamless flow.
Contas Australia's leading airline takes a broad look at the flow of travel and has invested
heavily in optimizing every detail of the customer journey. This begins with the core
airline business. Contas personalizes the booking, check-in, in-launch and in-flight
experience. For example, its app makes real-time recommendations according to where
the passenger is, such as how to check-in most efficiently, what time to leave for the
airport and the best route to take.

The airline has also thought beyond travel and for that it has built a loyalty ecosystem
across categories with hundreds of partners such as Wolfsworth, the leading Australian
grocer, Hilton, Avis, eBay and major Australian banks, enabling its customers to earn and
spend points in novel ways according to their preferences. So now you see that the
airline is doing what all in order to engage the customers and make their experience
worthwhile. This media analytics and research service Red Planet helps Contas and many
of its partners combine offline and online behavioral data with media buying to target ad
campaigns. Contas has also used its data to launch new businesses. For example, it
designed an app with which customers can earn points for healthful habits such as taking
a certain number of steps each day or working out regularly.

And to unlock the points, customers are invited to sign up for the airline's new health
insurance business. So that is an airline getting into health insurance. This app also
enables the company to cross-sell travel and travel to other countries. The airline is also
using the app to launch new businesses. To orchestrate communications about these
offerings, Contas built a marketing messaging platform that leverages AI and a library of
personalized content to deliver the right message through the right channel to each
customer.

The third building block is to activate the experience across channels. So, Starbuck is
famous for its personalization across channels. Its app delivers gamified offers based on
individual preferences and behavior. Its paid digital media ads are highly targeted and its
in-store experience including digital menus in the drive-thru that change according to
weather, local customer preferences and inventory. Although many smaller restaurants
chain struggle to compete with Starbuck's level of personalization, Sweetgreen which has
only 140 stores worldwide.

While Starbuck has 33,000 plus builds its cross-channel experience with data and data in
mind. It launched a best-in-class app that makes it easier to create a customer's salad and
pick it up on or have it delivered. It uses the app to roll out new digital menus and deliver
personalized offers for customers and it allows in-store customers to pay by phone. The
app enabled Sweetgreen to surpass Starbuck's percentage of digital engagement in 2021
with 68% of Sweetgreen sales coming from digital channels compared with only 52 for
Starbuck stores in the United States. The fourth is, fulfill according to the customer's
context.

Huge retailers like Kroger and Tesco have large data and analytics teams that build
algorithms for use in engaging customers in ways that are most appealing to them.
Kroger's and Tesco's analytics arms, 84.51 degree and Dunnhumby respectively run
hundreds of propensity models to decide which personalized promotion to offer to which
customers. The mid-sized grocery chain, Giant Eagle has also entered this space. It is
partnering with formation and innovative software-aided service tech company to achieve
the same level of personalization in targeting its promotions.

The grocer has gamified the shopping experience, rewarding its customers with loyalty
points whenever they complete certain steps arranged via its app. For example, new
customers might be invited to complete a weekly shop challenge that encourages them to
come to the store once a week during a specific month to earn extra points at Giant
Eagle’s Fuel Perks Plus program, good for free gasoline or discounted groceries. Loyal
and long-term customers might receive points for shopping in a new category that,
judging from similar customers' profile, probably interests them, such as chocolates. The
fifth is to test relentlessly. StitchFix is a digital native that encourages and incentivizes
its team to run hundreds of experiments every month, fully expecting a third of them to
fail.

It feeds the data from these experiments into its intelligent experience engine to inform
the next best action. It also asks customers for data directly. So, these are some, this is the
logo of StitchFix, and these are some pictures from that. StitchFix Style Shuffle is an
interface that new subscribers can engage with when they sign up for the service.
Customers' wives write for items they like and left for ones they don't, giving StitchFix a
clear view of their personal taste and style.
The company's algorithm then extrapolates that data from a few items to thousands of
SKUs to help craft the monthly selection of apparel delivered to the customer's home.
The sixth building block is Honing the Craft. Today, leaders are going further to indoor
teams with even greater responsibility for leveraging data. The teams essentially serve as
product managers dedicated to continually improving end-to-end customers' interactions.
To begin the process we have described, you should ask, what experience do we want to
revolutionize and how can we build an intelligent engine to achieve our goal? Once
decided on the answers, research a few customers' records in your CRM and marketing
automation platforms to determine whether we have captured all the relevant data needed
to power more valuable experiences.

Most CEOs and their C-suite colleagues claim to recognize the importance of the
customer experience, but we often see more talk than action and that must change. Every
company needs an explicit strategy for building an intelligent experience engine which
can align the organization towards using AI, personalization and agile processes to build
deeper, more enduring brand loyalty. The next thing that we will understand is Hyper-
Personalization using AI. So now let us understand how customers' expectations are
changing. Marketing leaders must recognize that providing a superior customer
experience is not only expected but a competitive advantage.

Experience is critical to capture new customers and retain existing ones in today's
crowded digital landscape while also maintaining or increasing the share of wallet.
Customer experience is becoming even more important given that increased globalization
offers customers an abundance of choices. However, to meet and exceed evolving
customer expectations, brand must use data and analytics to better understand their
customers and harness the power of AI to create authentic interactions and the hyper-
personalized experience that customers now expects. That is one. The second is the
competition is becoming more advanced now.

Traditional business models are becoming increasingly outdated. Companies are

consistently challenged by new entrants to the market that offer lower price points or
more innovative experiences. Companies like Amazon, Facebook and Google are leading
the charge through their use of rich customer databases and personalized
recommendation solutions. Digital marketing is also becoming a crowded sector leading
to increased spending and hopes of reaching customers. When faced with such a keenly
competitive landscape, AI-powered marketing offers the ability to make a more
meaningful impression on your customers with the limited exposure they may have to a
brand's message, allowing them to forge deeper bonds and excel in the marketplace.

The third is that data is increasingly available. Currently customer data was collected
across outdated and disconnected systems and limited to call centers and point of sale, so
that data was not of much use. However, increased digitization coupled with advanced
data technologies allows organizations to use their propriety and third-party data to create
detailed pictures of customers and gain a deeper understanding of their preferences and
behaviors. AI-powered marketing technology can be used to sift through vast amounts of
information in real time and make insight-driven decisions on the type of interactions to
have with each customer. Organizations must begin to effectively use available data and
AI technology to accelerate results and offer hyper-personalized experiences to attract
and retain loyal customers.

In order to excel in today's marketplace, organizations should adopt a hyper-personalized

marketing strategy by exploring ways data, advanced analytics and AI technology can be
embedded throughout the customer journey and generate new levels of industry-specific
customer insights and actions. Now we will understand what hyper-personalization is.
Hyper-personalization is the most advanced way brands can tailor their marketing to
individual customers. It is done by creating custom and targeted experiences through the
use of data, analytics, AI and automation. Through hyper-personalization, companies can
send highly contextualized communication to specific customers at the right place and
time and through the right channel.

As digital marketing becomes more competitive, hyper-personalized marketing provides

the opportunity for organizations to meaningfully engage customers, deepen existing
relationships and build new ones and improve the customer experience. Implementing
this type of strategy not only increases customer satisfaction but also drives brand loyalty,
willingness to spend and overall marketing effectiveness. So, these were the CMO's
goals. First was to increase profit, the second was to innovate and then optimize, the third
was to improve experience and the fourth was make data-driven decisions.

So these are the CMO's goals. Now hyper-personalized marketing can help a CMO by
maximizing the revenues, reducing the cost and elevating customer experiences. So, this
is how this hyper-personalization can convert the CMO's goal. So, the CMO's goal, the
first is to drive profit, drive business growth through reimagined platform and service
design, plan for delivery of top-line revenues and ROI, lead the shift from cost center to
profit center and integrate product service development with marketing and sales. The
second is to innovate and optimize, employ new marketing strategies and tools to drive
growth and reduce costs. Cost will include lower customer acquisition and the retention
cost through personalized product targeting and campaigns.

The third CMO goal is to improve experience, create and ensure a consistent,
personalized and memorable brand experience across all media through a deeper
understanding of customer needs and relevant touchpoints. The fourth is to make data-
driven decisions, deliver concrete and understandable marketing metrics and analysis for
data-driven decision making and risk management. Now hyper-personalized marketing
can help by one, maximizing the revenue by way of data-driven content generation,
detailed product targeting, next best action or recommendation engine and individualized
or dynamic pricing. It can help in reducing costs by reducing customer acquisition and
retention cost and workflow automation. It can help elevate customer experience by way
of in-movement customer journey, 24 by 7 personalized customer service, real-type
customer segmentation and dynamic landing pages and best sites.

Now let us look at ways to collect data for hyper-personalization. So, we will discuss
three ways to collect the right kinds of data for a hyper-personalization campaign. First is
social listening, second is attribute analysis and the third is behavioral segmentation and
event sequence analysis. What is social listening? Social listening is the act of using
publicly available data points on social media and social media platforms to personalize
marketing content that you send to your customers. You can use hashtags on Twitter and
Instagram, comment sections on Reddit, viral videos on YouTube, popular posts that
customers share on Facebook and so much more.

You are only limited by the data points that each social media website and platform
provide for public consumption. Social listening is vital for businesses that occupy the
public sphere as well as businesses that are politically or socially aligned with specific
causes. On a regular basis, you can analyze data points from your choice of social media
websites and platforms using social listening, considering the public and social response
to your cause and then incorporate these developments into tailored hyper-personalized
marketing campaigns. All of this information is readily available and easily collected
using the many software options that are available on the market. The key to social
listening is to define who your customers are and what they want.

This is a similar process to creating customer profiles but much more targeted and using
readily accessible data points. You may even be able to isolate pain points that the
customers experience during the shopping process. Pain points are the areas of frustration
that customers experience when using an online shopping platform. An example of a pain
point would be the need to login to save a shopping cart locally. If a user has forgotten
his or her login credentials and is on a public computer, this experience is frustrating and
may cost the company a sale.

Eliminating these pain points will help a business convert more sales and provide an
easier purchasing process for future customers. The quicker and the simpler the process is
to purchase items online, the more leads you will nurture. Example of social listening is
social listening through the lens of hyper-personalization enables the company to directly
cater marketing materials to a customer. If a specific customer posts on social media
about an upcoming party they are having for a big sports event, a hyper-personalized ad
may feature a popular pizza restaurant advertising to them closer to the event. This could
surface through paid advertisement on website and social media platforms or it may take
it a step further and suggest that the customer needs to have a few large pizzas for the
upcoming game.

When it comes to hyper-personalization, the possibilities really are endless due to how
deep and personal these advertisements can get. The second is Attribute Analysis.
Attribute Analysis is a marketing tool that will help you break down associations between
individuals in your customer base. Using Attribute Analysis to research your customer's
data will help you consider new ways to personalize the shopping experience for your
customers both as individuals and collectively as a pool. With Attribute Analysis, you
plot down every single attribute that you can find concerning your customer.

You then use these attributes to tailor the shopping experience based on your marketing
goals concerning hyper-personalization. A few examples of these attributes include 1.
Activities 2. Interests 3. Opinions 4. Behavior 5. Values.

One example of Attribute Analysis is Dating Site Questionnaires which use a similar
process to match to individual based on commonly shared data points like whether an
individual smokes or has a pet. Using Attribute Analysis, these companies match you
based on the data points they collect about you. You can extrapolate this technique and
use it in your own marketing campaigns to enable you to reach your target audience. 5.
Behavioral Segmentation and Event Sequence Analysis which can be a part of a broader
behavioral segmentation strategy that identifies users' type based on their browsing
behavior.

Is the process of collecting data from a customer while he or she is purchasing a product
from you? That could include everything from the use of a coupon, whether or not the
customer is logged in, how quickly a customer made a purchase against the average
purchase time of all customers and so forth. Event Sequence Analysis can indirectly
benefit customers, allowing your hyper-personalized marketing plans to become even
more effective if implemented correctly. And Online Store that requires members to login
to save their cart rather than use locally stored cookies to save the cart and then apply
their membership data at the checkout stage of the purchase process will be slower and
more cumbersome to use. If you notice that members are abandoning their online
shopping carts or not making as many purchases as you expected, simplifying the
shopping process based on these data points can lead to increased revenues for your
company. Traditional personalization strategies saw companies send cart abandonment
emails while a more hyper-personalized approach revolves around discussing the specific
items they have abandoned, offering incentives for those items and creating limited time
offers that cater to the customers.

So, to conclude in this module, we have continued explaining the Intelligent Experience
Engine and building it with 6 pivotal practices. Then we understood about the changing
scenarios that are leading to hyper-personalization. Then we have understood what hyper-
personalization itself is and then we have also discussed the CMO goals and the benefits
when these those are transformed to hyper-personalized marketing. Finally, we have
discussed ways in which data can be collected for hyper-personalized campaigns. And
these are the 8 sources from which the material for this module was taken. Thank you.
 AI in Marketing
Prof. Zillur Rahman
Department of Management Studies
Indian Institute of Technology, Roorkee
Week 7
Lec 35-Personalization and hyper-personalization Using AI-III

Welcome to this NPTEL online certification course on artificial intelligence in marketing,

and now we will talk about module 35. So, we are talking about personalization and
hyper-personalization using AI. So we have continued through module 33 and 34 with
that and now we will continue in this module also with the same topic, so that is part
three, and this is the module overview. The first is how hyper-personalization is different
from personalization. The second is hyper-personalization journey throughout the
customer flowchart. The third is what are the preference models and the factors to update
the model. The fourth is the key elements in the system architecture of hyper
personalization and the fifth is how does hyper personalization looks like for some
renowned companies and example of its wins and loss. Hyper personalization goes
beyond segmentation. While segmentation creates customer growths based on shared
likes and dislikes and activities, hyper personalization drills down to minute differences
which can be used to target customers at the individual level. So now we are talking of
this individual level and not several individuals.

Traditionally organizations have used customer segmentation as part of their marketing

strategy in attempt to ensure customer receive relevant communications and offers but
struggle to achieve deeper levels of personalization through this tried and true method.
While increasing segmentation efforts appears to be a good approach, it will not result in
the best ROI or maximize program effectiveness. AI-powered hyper personalization
delivers optimal results by allowing companies to tailor their marketing efforts at the
individual level by using data gathered on that specific customer. For example
personalized product recommendations or unique discounts can be shared using unique
customer data such as psychographics or real-time engagement with your brands.

This segment of one approach allows us to optimize whom we target with key messages
and offers through the most relevant and appropriate channels. Implementing this
strategy not only increases customer satisfaction but also drives brand loyalty,
willingness to spend and overall marketing effectiveness. Hyper personalization can be
achieved in various degrees ranging from recommendation engines to connecting online
and offline sales channels and from predicting customer preferences to developing
tailored product or pricing. This level of personalization cannot be achieved through the
implementation of a single business case. Rather it is a holistic marketing strategy that
fundamentally changes the way organizations interact with customers and should be
treated as an evolving and maturing practice that is embedded throughout the customer
journey and part of every marketing campaign.

As organizations advance their personalization efforts, they can expect deeper

relationships to develop with existing customers while also attracting new customers.
What are the key differences between personalization and hyper personalization? If
personalization is advertising back to school supplies for individuals who purchase
soccer balls online in August, hyper personalization is advertising these same school
supplies with optimized advertising based on the location the customer purchased, the
time of the purchase, whether or not the customer used a credit card, whether the
customer mentioned soccer and related activities on social media and more. Hyper
personalization is more involved, more complex and more effective than personalization.
Traditional personalization deploys profiling techniques to make assumptions about the
user based on certain traits, allowing the marketers to tailor messaging, products or
services based on these traits. But this is far less detailed approach than using a specific
customer history and real-time context to truly understand the user's needs and intent.

Through hyper personalization, brands can identify the subtle detail about the customers
that traditional personalization and profiling fails to catch, which in turn helps them to
provide highly targeted and personalized products, services, promotions and content.
Now this is the personalization maturity curve. On the x-axis, we have personalization
maturity and on the y-axis, we have revenues and they are increasing. So the revenues
are increasing as the personalization becomes more mature. So it starts with lower
revenues and we have single message mailing.

Then comes field insertions. Now you see that with as we move on, on this axis, the
revenues are increasing, rule based segmentation. Then again now behavioral
recommendations, omni-channel optimization and then comes predictive personalization.
So at the highest degree of personalization and that gives you the maximum revenues
which is called as predictive personalization that is used by Amazon and Netflix and
Starbucks. Hyper personalization can be applied throughout the customer journey from
attracting customers with personalized web pages and dynamic pricing to provide
personalized service after the purchase.

Unlike with mass media where marketers can only assume which customer type of
segment may view and identify with a specific advertisement, hyper personalization
advertising uses the same platform and underlying data to present one of a multitude of
targeted offers based on who is viewing the offer. Cadbury made use of customer
information and photos to create a hyper personalized video ad. This campaign used
customer data including age, location and likes provided via the customer's Facebook
page after receiving consent and it resulted in increased click through rates of 65% and
conversion rate of 33.6%. Organizations like Amazon continue to experiment with
personalization after the advertising phase seeking to increase sales conversion through
the use of recommendation engines that serves customers with the exact product they are
looking for.

While this experience is so seamless that customers may not even realize
personalization is occurring, customers now expect brand to act like Amazon and predict
the product that fits their needs. Now this is hyper personalization customer journey
flow chart. So it is start with custom and relevant advertising that leads to unique lending
pages for each customers. Then this information goes to the recommendation engines
and they make recommendation according to the uniqueness of particular customer.
This dynamic pricing and offers the prices change and offers are made according to the
customer's requirement.

Then it goes to the chatbots. Then comes the omni-channel customer service across all
channels, pre-populated applications, real time product notification and royalty programs
and re-engagement. So that all those things that you have seen in the case of Qantas
airlines. So the first is custom and relevant advertising. Targeting customers with
advertisements that are unique by either including relevant products or customer
information.

The second is unique lending pages using queues on where customers are coming from,
past visits, geographic data and preferences to choose what is being presented to them.
The third is the recommendation engines providing content, product or service
recommendations that are tailored to the individual customers needs or wants. The
fourth is dynamic prices and offers. Changing the offer, promotion or price customers
are served based on their propensity to convert. The next is service chatbots using
conversational AI technology that learns from customer behavior and delivers
personalized services that can answer specific questions and concerns in real time.

The sixth is omni-channel customer service using databases and AI technology to

recognize and connect customers from both online and offline shopping channels. The
seventh is pre-populated applications where existing customer data to pre-populate any
document, processes or applications that may be required. The eighth is real time product
notification providing customers updates on the status of product shipments, promotions
or refills based on their purchase history. The last is the loyalty program and engagement
using customer purchases, micro-segmentation, geospatial data to send highly
contextualized offers and messages. Next comes hyper-personalization with the use of
preference models.

Preference models are an evolved form of choice models. However they don't just
model a customer choice between the product A or B but go much further and model the
customer preference both stated and inferred using deep learning. A preference model
learns to model a user stated and inferred preferences from actions and information about
the user. The best way to think of it is the model imitates the conscious and subconscious
of the user in order to best replicate how it would behave under any circumstances. The
model is the aggregate of everything that has been learned about the user and is used to
predict what a user may do given a circumstance such as classic choice between product
A and B.

Or it could be more advanced use case like how to generate a product image for an
advertisement that will most appeal to and resonate with the customer. A pre-trained
preference model may be fine-tuned on a dataset from a class that most closely
resembles the consumer based on conventional classification methods for example
profiles for similar demographics, interests and preferences. The preference model is an
active learner so it is constantly updating its weights that is the internal store of
information based on the interaction it has with the customers or supplemental
information provided to the model about the consumer. Some of the factors that update
the model are 1. Demographics, learned personal traits and characteristics such as your
age, gender, race or personality type used initially to set a model or to correct the model
when it drifts.

A special attention needs to be paid to ensure no bias or stereotypes are introduced with
this sensitive personal data. The second is location and activity data, information from
location and activity data such as your movements and the places you visit. Behavioral
patterns, learned behavioral patterns and habits such as the times of the day you are most
active or the types of activities you engage in regularly less about what you say and more
about what you do. So that is what behavioral patterns are. Personal preferences, these
are both stated and inferred.

Stated and inferred personal preferences and are learned by analyzing past interactions
such as your search history or purchase history, the content of conversation with a
personal assistant or chatbots or any stated or inferred preference. The next is the state
of mind, mood and sentiments. Learnt sentiments and emotions by analyzing your
written or spoken language through a chatbot or personal assistant dialogue, a phone call
or public social media post. Personal traits and characteristics, learned personal traits and
characteristics such as your age, gender and personality type by analyzing various data
sources and making predictions based on statistical pattern. Then, responses to
recommendations, the model can be corrected by assessing the success of past
recommendation in order to perfect the model for future predictions.

Conversational patterns, through analyzing the way someone speaks or writes, the
model can adapt to carry out this conversational style in the generation of speech and
text. Outside model behavior, any data not from direct user interaction that comes from
third party and does not fit into one of the above categories. The key elements, hyper
personalization is defined by two key elements. One is understanding the customers, the
ability to model customer preferences. And second is then to personalize the content, the
ability to generate content that will most resonate with the consumers.

So, the more you know about the consumer preferences, more individualistic the content
will be. Now, how do you understand the customers? Understanding the customer
preferences achieved through the use of something like a preference model to best model
the consumer. The model includes immutable attributes like demographic information,
long term attributes like state of mind, personality, mannerism and behavior, short term
attributes like mood and current locations and historical attributes like travel patterns and
search histories to name a few. So we are looking at long term attributes, short term
attributes and historical attributes.

Personalizing the content. Conventional personalization techniques like content

recommendation and ad serving are coupled with generative techniques like dynamic
generation of highly personalized marketing material or generation or dynamic
modification of multimedia like in movie branding and video style transfer where the
video is altered to match the preferred style of the viewer. The next comes system
architecture. Deployment architectures for hyper personalization platforms can vary and
can be very complicated as per requirements. But a very simplistic architecture of a
hyper personalization platform will be explained in the next slide.

This is that simplistic model. So this is a large language model. Then we have image
generation model and then we have a preference model. Here we have retrieval model
endpoints. Now this goes through the conversational AI, image generation, consumer
models and prediction, forward channel delivery and return channel that is feedback.
And then these are external user preferences are included in this preference model.

So there are two channels in interacting with the customer. The forward channel that is
delivery and in the return channel that is the feedback. The forward channel is the
delivery of recommendations, predictions and dynamic content generation to the
consumer as well as conversational responses when implemented with a chatbot,
personal assistant or search engine. It is a communication channel back to the consumer.
The reverse process is the way by which feedback is gathered from the consumer and
processed by the hyper personalization system including the storage of data.

If implemented in a conversational manner or using some form of search, the input into
the system is used to respond to the query as well as to understand the consumer. Other
inputs beyond text and speech may go through these channels including location data
and other behavioral data. Actively learning is an important feature of hyper
personalization. Behaviors mood and interest can be temporal or can begin and end at
any point in time. Therefore this active learning becomes important.

The input into the system is used to understand the consumer's preferences and
continually update the preference model to keep accurate account of their current
interests, mood and so forth. External data that updates the model can come from
external sources and that data can be personal data like social media post or class or
general data like current world events or trends. This data also updates the preference
model in real time to ensure that the system is providing the best recommendations and
content for that specific moment in time. What does hyper personalization look like?
Amazon hyper personalization, hyper personalized customer experience. Customer
receive highly contextualized emails with personalized product recommendations based
on customer demographics, psychographics and previous purchase and view history.

The channel is email. Approach is user recommendation algorithm called item to item
collaborative filtering to suggest products based on key data points to create user profiles
and craft a highly contextualized email for the shopper. Data considered customer
demographic, search query, average, time spent on searches, past purchase history, brand
affinity, category browsing habits, time of past purchases, average spend amount.
Outcome, the product recommendations engine generates over 35% conversation by
creating unique hyper personalize experience for each customer. Netflix, how it hyper
personalizes customer experience? Customer receives a highly contextualized and
individual experiences starting from a home page that is based on their past viewing
history and that uses viewing habits to recommend content. The channel is email and
push notifications.

Approach, Netflix user algorithm to predict content that users will want to see. It
combines behavioral attributes with predictive learning to send 103 million users unique
movie and show recommendations so as to increase engagement and loyalty. What data
are Netflix considers? Products of customer behavior including viewing history, ratings,
viewing times, preferred devices, viewing durations, movie information including title,
genres, categories, actors, release years, etc. Members with similar taste and
preferences. The outcome of this is Netflix recommendation engine has been critical to
customers retention as 80% of the users flow through on a recommendation and only
20% search for content.

Now let us look at the example of Starbucks. So how does it provides the hyper
personalize customer experience? So customers receive the personalized in-app
experience with real-time offers based on their preferences, activities and past purchases.
The channel is push and in-app notification. The approach is as part of its digital
flywheel strategy. Starbucks uses a data-driven AI algorithm to send over 400,000
variants of hyper personalized messages that is food beverage offers to their customers
and promote unique and compelling offers for each specific member.

The data considered is the contextual data including location data, geospatial,
demographics and traffic. Customer preferences, customer activity and past purchases.
So all these are this is all the data that they use. And the outcome is marketing campaign
effectiveness and incremental revenue via offer redemption increased threefold with an
estimated 25% of total transactions being conducted via the mobile app. Now we will
look at the case study of win and loss.

So targets hyper personalization mistake. Andrew Poll, a statistician working for Target
successfully determined whether a Target customer was pregnant even before she knew
herself based on her personal shopping history. This issue arose when Target was
anything but subtle when applying this data. Once Poll was able to determine which
female Target shopper may have been pregnant, Target shared their data points with
marketers. In this case, the father of the teenage girl complained to Target that his
daughter was receiving achievements for maternity clothing, nursery furniture and
pictures of smiling infants. At that point in time, the father and the daughter did not
know she was pregnant and Target's marketing campaign shocked them.

Although Target was using a form of personalization with good intentions to create a
better, more personalized shopping experience, Target did not consider the issue of
applying their data to all female shoppers. If Target had applied their data to only
women over the age of 21, it is likely that this type of hyper personalization would have
been more well received. So we can easily contrast this scenario with an example from
Amazon where they used better hyper personalization practices. But how did Amazon
win with hyper personalization? A good example of hyper personalization can be viewed
in Amazon's suggested item that you see on product pages and during the checkout
process. Amazon locks the products that you see, even if you do not purchase them.

Based on your browsing history, not just your purchasing history, Amazon recommends
similar products that you may be interested in based on what other customers with
similar viewing habits have seen. Over time, the end result is a highly tailored, highly
personalized shopping experience that Amazon competitors cannot match. 35% of
Amazon customers purchase items that are recommended to them. These
recommendations are automated using predictive modeling to consider which product
customers will be interested in based on their previous purchase history and the data
collected from the social media. Whether you agree with the way the data is collected,
this is an example of hyper personalization done right and also implemented correctly.

So in order to conclude in this module, we have understood how hyper personalization

is different from personalization. We have also discussed the steps in a hyper
personalization customer journey flowchart. Then we have studied about preference
models and how to update it. We have also gone through the key elements as well as the
system architecture of hyper personalization. Finally, we have understood how hyper
personalization looks for some famous brands and a case study of hyper personalization
win and loss.

And these are the nine sources from which the material for this module was taken.
Thank you.