Principles of marketing group 1 class cc01 application of artificial intelligence in marketing

However, it is conceivable that Artificial Intelligence Al in marketing refers to the use of advanced technologies such as machine learning, natural language processing NLP, and predicti

VIETNAM NATIONAL UNIVERSITY — HO CHI MINH CITY

HO CHI MINH CITY UNIVERSITY OF TECHNOLOGY

APPLICATION OF ARTIFICIAL INTELLIGENCE IN MARKETING

Supervisors: Dr Bui Huy Hai Bich

Nguyễn Vũ Hải Đăng 2052446

Phó Đức Gia Bảo 2052401

Đào Hiền Mai 1952833

Nguyễn Quốc Thanh 1952976

Table of contents

CHAPTER 1: OVERVIEW OF ARTIFICLAL INTELLIGENCE IN MARKETING 2

1 Definition and classification «2.0.0.0 ieee eee ố ỐỐ 2

Am 2

M9 nh 6 6 dA4 2

2 Reasons for AI Marketing Trends 2.0.00 ete 3

2.1 Benefits of AI in Marketing ốc inne 3 2.2 Potential Applications of AI in Marketing: - ST SH xxx nen 4 2.3 Future Irends in AI and Marketing: . -c ST SH HH kh reh 5 E0 áo v0 cố ốc ee 6

` 9

CHAPTER 2: CASE ANALYZING TQ nhàn nh nh KT kh nh 11 1 Amazon’s Recommendation engine background - xxx xxx xxx nreh 11 2 Algorithm of Amazon's recommendation €nØ11€ +5 5s xxx xe ờ 11 2.1 — Item-Item collaborative filtering ằ 12

"vốn cốc ĐS 14

2.3 Bandit and causal Inference in recommendatiONs se 14 3 The va ốc on cố 6 17

4, Evaluation of the effectiveness by SWOT analysis ccc cece eee e teeter tere eeeenaeaeaes 19 Thi oan Ố 21 SUMARY 157 c aaa 22

REFERENCES cQ nh nh nọ TT KT in TK kh xrh 23

CHAPTER 1: OVERVIEW OF ARTIFICIAL INTELLIGENCE IN MARKETING

1 Definition and classification

1.1 Definition

There is no definitive definition of Al in marketing from a leading organization in marketing that is widely recognized However, it is conceivable that Artificial Intelligence (Al) in marketing refers to the use of advanced technologies such as machine learning, natural language processing (NLP), and predictive analytics to automate and optimize various marketing tasks and processes Al can help marketers to understand consumer behavior and preferences, personalize marketing messages, and deliver a more engaging customer experience

Marketers can use Al to evaluate vast amounts of data, such as customer demographics, purchase history, and online behavior, to acquire insights into the needs and preferences of their target audience These data can be used to develop individualized and targeted marketing campaigns that increase customer engagement and conversion rates

Besides, with the help of Al, marketers may increase the effectiveness and efficiency of their marketing initiatives while also saving time and resources in many tasks by automation tools and platforms like content creation, email marketing, social media management, and customer support

1.2 Classification

In general, there are 3 main type of Al that may change in marketing process:

Analytical Al operates by analyzing large datasets to identify patterns, trends, and insights

It uses statistical algorithms and machine learning techniques to build predictive models that can forecast future outcomes and make data-driven decisions Analytical Al can be used for customer segmentation, predictive analytics, and marketing attribution modeling The process involves data collection, preprocessing, feature selection, model training, and evaluation Analytical Al can help marketers gain a deeper understanding of their customers' behavior, preferences, and needs, allowing them to create more personalized and targeted marketing campaigns

Generative Al is a type of Al that creates new data similar to the original by using a dataset

to train a model, optimizing it, generating new content, and evaluating the results It is used for text generation, image and video synthesis, and music composition Generative Al has many applications in marketing, including content creation, personalized product recommendations, and dynamic pricing Its ability to generate new and diverse content is becoming increasingly important in today's digital age, where content is king and engaging customers is critical to success Conversational Al operates by using natural language processing (NLP) and machine learning algorithms to enable human-like interactions between humans and machines It includes chatbots, virtual assistants, and voice assistants that can understand and respond to human queries

and commands Conversational Al uses deep learning techniques to analyze and learn from user interactions to improve its responses over time It is used for customer service, sales, and marketing, enabling brands to engage with their customers on various platforms such as social media, messaging apps, and websites The operating principles of Conversational Al involve NLP, machine learning, and dialog management, which enable it to understand, process, and respond to natural language inputs in a human-like way

2 Reasons for Al Marketing Trends

Artificial Intelligence (Al) is a rapidly developing technology that has had a significant impact on various industries One industry that has seen a significant transformation due to Al is marketing since it have outstanding benefits, potential applications, and future trends

2.1 Benefits of Al in Marketing

One of the primary benefits of Al in marketing is its ability to analyze vast amounts of data quickly and efficiently Al algorithms can process data at a speed that would be impossible for humans to match, allowing marketers to gain insights into customer behavior, preferences, and

patterns With this information, marketers can create more targeted and effective marketing

campaigns that reach the right people with the right message at the right time

However, leveraging artificial intelligence for content creation goes beyond simply relying

on ChatGPT to generate extensive amounts of text and publishing it directly on your blog For example, ClickUp, a project management solution, incorporates Surfer SEO's natural language processing Al tools and machine learning technology for various purposes:

Discovering opportunities to optimize content: ClickUp utilizes Surfer SEO's Al tools to identify areas where their content can be improved for better performance

Determining relevant keywords and their usage: Through Surfer SEO's Al capabilities,

ClickUp gains insights into the appropriate keywords to include in their articles, along with their

optimal density for search engine optimization

Obtaining guidance on ideal article structure: ClickUp taps into Surfer SEO's machine learning technology to understand the recommended structure for their articles, including factors such as the number of images to include and the length of subheadings

° 132 words in h2 to h6 ˆ Hide details

www.indeed.com foirygodboss.com troqq.com www.wgu.edu www lifehock.org

#3 in Google #4 in Google #5 in Google #7 in Googie #10 in Google L] You Competitors Suggested ronge

Figure 1 1 ClickUp use NLP AI to boost blog trafic by 85% Source: Surfer SEO

The outcomes have been remarkable Since implementing Surfer SEO, ClickUp has experienced an impressive 85% increase in organic, non-branded traffic Additionally, the organization has achieved enhanced efficiency in content production, completing more than 130 optimizations and publishing over 150 blogs

By adopting Surfer SEO's Al tools and machine learning technology, ClickUp has witnessed significant improvements in their content strategy, leading to enhanced organic traffic and increased productivity in content creation

2.2 Potential Applications of Al in Marketing

Al has several potential applications in marketing, including predictive analytics, chatbots, and programmatic advertising Predictive analytics involves using Al algorithms to analyze customer data and predict future behavior or trends This can help marketers to identify opportunities for upselling or cross-selling and create more targeted campaigns

Al-generated images have often received criticism, primarily due to their imperfect depiction, particularly in areas like hand-drawn elements However, Heinz, in collaboration with

marketing agency Rethink Ideas, has taken a different approach They recently introduced "the first-ever ad campaign with visuals created entirely by artificial intelligence." This innovative campaign marks a significant step in leveraging Al to generate visuals for advertising purposes, showcasing the potential for Al to contribute to the creative aspects of marketing

s ma ~ ” “ THIS TS WHAT - THES IS WHAT z

“KETCHUP” Gar @KETCHUP’) KETCHUP’)

LOOKS UKE TO Al LOOKS UKE TO Al LOOKS UKE TO A.l

What does Al think ketchup looks like?

Figure 1 2 Heinz Launches Its First Ad Campaign With Entirely Al-Generated Images

Source: Rethink Ideas 2.3 Future Trends in Al and Marketing

Al is expected to continue transforming the marketing industry in the coming years One trend that is expected to become increasingly popular is hyper-personalization goes beyond traditional personalization by creating experiences that are uniquely tailored to each individual customer Al algorithms can analyze data in real-time, providing insights that can be used to create hyper-personalized experiences that are relevant and meaningful to each customer Hyper- personalization not only improves the customer experience but can also drive business results Amazon pioneered personalized product recommendations using machine learning However, extending these capabilities to businesses running their sites on Amazon Web Services (AWS) has been challenging

Amazon launched Amazon Personalize in June 2019, extending its machine learning

technology to AWS customers This allows businesses to integrate personalized experiences and recommendations into their applications using the same powerful capabilities used on Amazon.com

Since its initial launch, the functionality of Amazon Personalize has been significantly enhanced It is now capable of delivering up to 50% improved recommendations for a wide range

of fast-changing product categories, such as books, movies, music, and news articles

Amazon Personalize Wentify feature Build feature store Customized

tomatically process and Personalization API

Provides Amazon Personalize

Amazon Personalize AP!

Stream use activity fre y

application 4 ì hyper-parameters aches

the Amazon Personalizi

API or JavaScript

Amazon Personalize Figure I 3 Amazon Launches Personalize Source: VentureBeat

Prominent brands like Domino's, Yamaha, Subway, and Zola are already leveraging Amazon Personalize for various purposes These include showcasing musical instruments and in- store catalogs, providing ingredient and flavor recommendations, as well as creating personalized style combinations for their customers By utilizing Amazon Personalize, these brands can offer tailored experiences and enhance customer satisfaction in their respective industries

3 Operating principles and application

In general, Al is divided into 3 main groups: Analytical Al, Conversational Al and Generative Al It is further divided into several smaller categories serving many different purposes Note, in some cases, they have to combine different types of Al to create a certain application The table below show common categories of Al in marketing:

Make recommendations or decisions based on a s¢

Prescriptive analytics predefined goals or constraints

Chatbots Simulate human conversation through text or voice

Conversationa Use speech recognition technology to understand |

Al Personal assistants respond to user voice commands

; Interact with customers through various channels, s Virtual agents as chat, voice, or email

Generative Used for image and video synthesis, allowing marke

Adversarial Networks to create new visuals for their campaigns

Generative Al Autoencoders Content personalization, by encoding user behavior

generating personalized content

Prescriptive analytics refers to the use of Al to make recommendations or decisions based

on a set of predefined goals or constraints In marketing, prescriptive Al can be used to optimize marketing budgets, determine the best channels or messaging for a given campaign, or personalize content based on user behavior

Example: Meta, Amazon, Google Ads.,

Marketing attribution uses machine learning algorithms to determine the impact of

different marketing channels on customer behavior This allows marketers to better allocate their

marketing budget and optimize their campaigns

Example: Adverity is a data management and marketing analytics platform that uses Al to help businesses track the performance of their marketing campaigns The platform integrates data from multiple sources, including advertising platfiorms, social media channels, and website analytics tools, to provide a unified view of campaign performance Adverity uses Al algorithms

to analyze this data and provide insights into which channels and campaigns are driving the most conversions

Emotional analytics is the use of Al to recognize, interpret, and respond to human

emotions In marketing, emotional Al can be used to analyze customer sentiment and feedback, or

to personalize messaging based on emotional cues such as tone of voice or facial expressions Example: Coca-Cola: In 2016, Coca-Cola created an Al-powered vending machine that used facial recognition to analyze customers’ emotions and recommend a drink based on their

7

mood The machine used Emotional AI to analyze customers' facial expressions and provide personalized recommendations

Sentiment analysis uses natural language processing (NLP) techniques to analyze customer feedback and determine their sentiment towards a brand or product This information can be used to identify customer pain points and improve the customer experience

Example: Hootsuite Insights: Hootsuite Insights is a social media monitoring tool that uses sentiment analysis of Al to help businesses understand how customers are reacting to their brand and products on social media The tool uses natural language processing (NLP) algorithms

to analyze social media conversations and identify whether the sentiment is positive, negative, or neutral

Chatbots can be used for a variety of customer support tasks, such as answering FAQs, resolving issues, and guiding users through complex processes

Example: Domino's Pizza uses a chatbot of Al called "Dom" to take pizza orders through Facebook Messenger Customers can use the chatbot to order pizzas, track their delivery, and get answers to common questions

Personal assistants use NLP and machine learning to understand user intent and provide personalized responses

Example: Starbucks uses an Al personal assistant called "Barista" to take voice and text- based orders from customers Customers can place orders through the Starbucks app or by using voice commands with Amazon's Alexa or Google Assistant

Virtual agents use NLP and machine learning to understand user intent and provide personalized support and recommendations, interact with customers through various channels, such as chat, voice, or email

Example: Autodesk uses an Al virtual agent called "AVA" to provide customer support for its software products The virtual agent can answer questions, provide troubleshooting tips, and help customers with technical issues

Generative Adversarial Networks can be used for a variety of tasks, such as image synthesis, text generation, and music composition They can be used for image and video synthesis, allowing marketers to create new visuals for their campaigns They are able to create unique, personalized experiences for their customers, generate new and innovative content, and stay ahead

of the competition

Example: Coca-Cola has used GANSs to create personalized labels for its "Share a Coke" campaign The company used GANs to generate unique designs for each name on its bottles, based

on customer data and insights

Adobe: Adobe is using GANs to help designers generate unique designs and artwork The company has developed a tool called "Project Scribbler" that allows designers to quickly create

8

new designs by sketching rough outlines that are then turned into polished, finished designs using GAN technology

Autoencoders: There are 2 common types of them: variational autoencoder (VAE) and denoising autoencoder (DAE):

VAE can be used for product recommendations, by encoding user preferences and generating recommendations based on similar encoded representations They can also be used for content personalization, by encoding user behavior and generating personalized content based on similar encoded representations

DEA removes interfering information from input data, which they often use for image, audio noise reduction, clean customer data, and anomaly detection

Example: Amazon has used autoencoders to create personalized recommendations for its customers The company's recommendation system uses a type of autoencoder called a variational autoencoder (VAE) to generate personalized recommendations based on customer data and purchase history

Airbnb: Airbnb has used autoencoders to improve the accuracy of its search and recommendation system The company's search and recommendation algorithm uses a type of autoencoder called a denoising autoencoder (DAE) to filter out noise and irrelevant information from its data

Deep reinforcement learning can be used for dynamic pricing, by training an agent to

optimize prices based on customer behavior and market conditions It can also be used for customer engagement, by training an agent to optimize messaging and incentives based on customer behavior and preferences While DRL has many potential applications in marketing, it is still a

relatively new area of research and there are few examples of companies that have implemented DRL specifically for marketing purposes

Example: Uber, which has been using DRL to optimize its surge pricing algorithm Surge

pricing is the practice of increasing prices during periods of high demand, such as during rush hour or during a major event Uber uses a DRL algorithm called the Multi-Agent Deep Deterministic Policy Gradient (MADDPG) to learn from past surge pricing data and make real- time pricing decisions that balance the need to maximize profits with the need to maintain customer satisfaction

4 Conclusion

To sum up, Al is being used in marketing for several reasons, including its ability to analyze data quickly and efficiently, personalize marketing messages and experiences, and automate repetitive tasks It has several potential applications, including predictive analytics, chatbots, and

programmatic advertising As Al continues to evolve, it is expected to transform the marketing

industry even further, with new trends emerging, such as the use of voice assistants and the

9

integration with emerging technologies like AR and VR In the light of wide range of benefits it offers to businesses that Artificial Intelligence has become a hot topic and indispensable part of Marketing

ARTIFICIAL INTELLIGENCE IN MARKETING eters Reporting Al Use

/ | 186%

|

Figure 1.1 Growth of Al in Marketing Source: Salesforce

Al in marketing is not a distant idea but a current reality Salesforce reports that the

adoption of Al among marketing leaders increased from 29% in 2018 to a significant 84% in 2020

As a result, the global value of Al in marketing is projected to rise from $12 billion in 2020 to an astounding $108 billion in 2028 This substantial growth demonstrates the increasing recognition

of Al's potential and its transformative impact on the marketing industry

CHAPTER 2: CASE ANALYZING

1 Amazon’s Recommendation engine background

Before Amazon applied the Recommendation Engine, the shopping experience on the site was primarily based on search functionality and traditional product categories Users had to manually navigate through the site to find products that interested them, and there was limited

ability to personalize the shopping experience

This approach had several limitations First, it was time-consuming and often frustrating for users to find products they were interested in, which could lead to a poor user experience and decreased sales Second, the lack of personalized recommendations meant that users may not have been aware of products that they would be interested in, which could limit sales opportunities for Amazon

To address these limitations, Amazon began to develop and apply the Recommendation Engine in the late 1990s, which allowed them to provide personalized product recommendations

to users based on their browsing and purchase history, as well as other behavioral data

Application of Recommendation Engine in e-commerce

Amazon applies the Recommendation Engine in various fields, including e-commerce, content streaming, and digital advertising In this presentation, we will focus on e-commerce field The main reason why Amazon applies the Recommendation Engine is to provide personalized recommendations to its customers, based on their purchase history, search history, and other behavioral data In e-commerce, the Amazon Recommendation Engine is used to provide product recommendations to customers based on their browsing and purchase history By analyzing customer behavior and identifying patterns, the recommendation engine can suggest products that

are likely to be of interest to individual customers This helps to improve the customer experience

on the platform and increase sales

2 Algorithm of Amazon’s recommendation engine

Based on what Amazon says about its recommendation system:

"We make recommendations based on your interests We examine the items you've purchased, items you've told us you own, and items you've rated We compare your activity on our

site with that of other customers, and using this comparison, recommend other items that may

interest you in Your Amazon Your recommendations change regularly, based on a number of factors, including when you purchase or rate a new item, and changes in the interests of other customers like you."

Generally, in e-commerce, the Amazon Recommendation Engine is used to provide product recommendations to customers based on their browsing and purchase history By analyzing customer behavior and identifying patterns, the recommendation engine can suggest

11