Hi, everyone. My name is Anand Kumar, and I'm a senior cloud engineer at SADA systems. And what I do for my for SADA system was to help customers in different verticals. to leverage artificial intelligence and machine learning language in their business and provide solutions based on, their, their business requirements. one of the industry that I work is, in retail. during this conference, I'm going to just talk about, how, machine learning will impact a professional, promotional strategies in one in the retail industry. So this is all about, the AI powered coupon personalization for, for customers, in the retail industry. so the topics that I'm going to cover is, some introduction about AI, which we all know, and, some promotional strategies in and around AI. what is AI powered coupon personalization, types of customers for personalization, how machine learning algorithms drive personalization, benefits of AI personalization, coupon personalization, some case studies, some real examples, that I can, I like to share and some successful implementation around those case studies. And, and also some challenges and ethical considerations, when using AI and, ML, and, future trends in, the AI powered promotions. let's get into the topics. First thing, introduction to AI and promotional, strategies. so to, before I start with, I'm sure most of you, might, think about discounts, promotions. offers, in the real time, world, especially in the retail. you go to any retail store to buy cloths or groceries or anything. the first thing that we noticed is about the discounts. when you enter the store, somebody will be standing outside the door and, welcoming you with, the promotion newsletter or paper, 50 percent off, 20 percent off. so they just wanted to, welcome you with, their offers and discounts. but do you know that there is a big strategy that is there? there could be a business strategy behind there for every of those, coupons or, the newsletters that you receive. What, the topic is, what the today's topic that I'm going to do, help you to understand is how this AI and ML is going to help these retail companies, retail, industry. as well as customer, to take the full advantage of, of the product or, or the discounts, that, you know, that the retail, the retail industry sales, the retail shop sales, or, the customer who's going to purchase, purchase that. So quickly start with, digital transformation, in retail. right now everything is data driven. in AI. Without the data, there was nothing. And, every, every moment, every, thing that you do in a retail or anywhere, it is all, the business is all driven by the data, strategies. So what is the AI role? So all these days we are using some data analytics, but now introducing the AI into the customer's data can help to give accurate personalization on the offers and resonate the customer. Individuals, individual consumers, how that particular offer that, that, that the retail shop gives, how that really benefits the customers. so we all know that, the retail, the e commerce market or the retail industry market is a 6. 3 trillion dollars, by end of this 2024. And, how, yeah, AI is going to personalize those offer, and crucial competitive, against these retail, companies. so let's get into the real topic, from the next slide. so what is AI powered coupon, coupon personalization means? To give you, to start with and give you an example, say, for example, you're going to Walmart or Amazon store or anywhere in the world, like even in India, there are big bazaar stores or Reliance stores, the moment you enter, there might be a desk or somebody would be standing and, giving you those, coupons, or offers, or you might get, some, letters, or you might get some messages to your phone, email that, hey, this is a festival offer, it's Christmas, it's New Year, if you purchase it can be like 20 percent offer, or if you buy these two, there is a third one will be an offer, and today's rate for, if you buy two shirts, there will be a pant, it's free, or it can be anything. But most of the time have you realized that, that those are not even necessary for me at the time that I wanted to purchase. for example, you might be, say you go to a retail store, There you wanted to buy a shampoo from a different company and a conditioner from a different company. Say you wanted a shampoo from L'Oreal and you wanted conditioner from Aveeno or some other brand. But, unless otherwise You buy both, there might be an offer, or if you're buying separate, there is no offer. and sometimes I don't want a shampoo at all, or I use something herbal, which I buy online or something. So you don't need that. Or, there might be another example where, I am, I love wine. I wanted to drink wine. Or some other preferences will be, they don't want to drink wine or, any alcohol or anything. But in the paper, it's a very common, collection of offers will be there. or say a pineapple. so some offers would say if you buy two pineapples, there will be another one, will be, on offer. But I don't need pineapple on for that particular day. Why would I need, that on a, on a winter time, pineapples are for summer, but the offers that they give you will be on winters because the retail store know that we have to give them offer since that has not been sold or there is an excessive amount of, produce is already there. Most of the time when I go, either I don't like that offer at that point of time, or I don't, I don't want it that, that, that particular time. what a, what a PowerCoupon Personalization does is, every retail store you go, there might be an account, they might ask your phone number, they might ask you for, the store, store customer number or something. they always collect the data. and, whenever you enter the bill, they might ask you, Hey, do you have the barcode? Do you have the, your customer? Do you want her to sign up for that retail? So everything is all data collected. We collect datas. rather than, So when I say data, the data is collected from your home number, your house address, or something that whenever you go, they know that it is you're purchasing. So think about a year worth of data, what all you have bought from that particular store, shirt, or you bought it from, from that particular retail, say it is a Walmart for our example, they store it in, in their database. So they know this particular customer. is planning to buy, or, he has bought this many number, of products from us and every time, or every month, he might have bought a particular set of, product. Say it is, it can be shampoo or it can be onions or it can be like, like shaving blades or creams or, or anything that. So what we are doing is, We are collecting all that particular data for that particular customer throughout the year, okay, or six months. And we are trying to personalize a coupon just for him. So rather than a very generalized discounts and offers, we want that customer to take leverage on what he has purchased on his history. Say, again, going back to the example, I love wine. if I wanted, they know from my every month I have bought this particular brand of wine. and say, okay, you're buying that brand of wine, by this time we are giving you an offer. If you buy two bottles of that wine, there might be another bottle of wine that is free. Or, think about another set of brand of wine. So if you buy that brand of wine, you might get an offer from this brand or another brand. you can just, personalizing or trying to understand the customer's behavior, based on what he has purchased all through these years and giving an offer would be more valuable for the customer and for the, and for the retail store to, hold that customer for a longer time. I bet that made, that really makes sense, right? how I achieve this is, we have to definitely make sure the customer is a member of that particular store. Say, if it is a Walmart, I become a Walmart member, and, I just make sure I collect the reasonable data, from different sources, from, apps that the customer use, that the, then the, the website the customer has used, or, or his location, just based on different parameters. we, we use, machine learning languages and models, to predict what would customer might buy. how can I just, be a value add? say for example, most of the customers, like if you, if you see the trend, if I go and buy a wine from a store, I have the tendency to buy a chips or a chicken, uh, it really makes sense, right? Hey, you buy a wine, so you get an offer for your Lays chips or something else or you know, there is an offer that if you buy a wine then there is a there's some discounts on the chicken or some party product. So you think about your product personalizing your discounts only for that particular customer and that discount might not be available for, the other customers, people are spending thousands and thousands of money on the advertisement and, and doing a personalization would be definitely makes sense for customers, customers. I personally allow it. I'm sure, everybody, if there is a personalized discount for that particular person, I feel like I am, more valuable for that particular store. And I love to go, to that particular store. okay, moving on, how I achieve this, how do I know the customer's bad. So there are various things as I mentioned before, purchase history, six months, a year, or the frequency, I would, I would definitely allow to collect at least three years of data, his age, his gender, because, I don't want it to give a wine to a less, an 18 year or, or a teenager who's not turned 18 or 21. And based on the browsing history that he might have used, hey, is this product available in Walmart? I just collect that I know, and very contextual, data, sometimes it is mostly like that particular location. I live in that location. I have three walmarts around, around my area within around 10 million years. I usually go to any one of them. You trying to understand your customers better, that really makes sense. And, think about some social media activities as well. because right now, when you hit enter on any of the website, you go to a particular website, that person really knows that what he's, is he doing. Say if I go to Walmart website and I'm just searching it. don't think that I, only I know, Probably walmart also knows that I am I'm searching for this particular product or And also they would also know that How many times I have clicked this website what all the pages that I have visited? So this is all is the data collection event. this is this is all getting into into a database where where it is just trying to get into a meaningful so You so try to understand, what is that particular customer's personalization? What is this personality basically, Say for example, if I am 50 years old or 60 years old and a person, if I am diabetic, say I'm searching for diabetic or anything, probably I won't, I won't need a discount on a chocolate or, or a cake. Say if an offer comes to me saying, you buy two cakes, there is a third one, why would I buy a cake where I am already suffering from diabetics? And I might ignore that, right? Ignore that particular discount. I might buy for another one, that is a chance, but the chances of that is pretty less than me having that, right? I hope you get it. basically trying to know their age, gender, particular location. This all just derives me or just puts me into that particular, category or a box where, okay, understand that particular person's behavior, personality on what he's, he may or may not buy. And based on that, produce that, give him that personalized offer. Okay. I hope you, you're getting it. And, it's pretty interesting. Okay. Then comes the technical part of it. How do I achieve this? Okay, say, I have, I went to Walmart, I signed up there for their membership or I'm just taking Walmart as an example, but really I have personally done this project for a different retail, retail store, and it's been an amazing success. in future you might see this trends going on with the most of the retail stores throughout the world. okay, let's talk, get in, get deeper into the machine learning algorithms. so these algorithms, these are some of the algorithms that can derive you. One is collaborative filtering. you're collecting from, different, users, different patterns. Say, for example, if a person of, of an age, age between 20 to 30, you particularly know by now that what his, what will be his eating patterns will be, what he would love, I know we all love ice creams, for example, but the amount of ice creams you consume in your later Between 10 to 25. It will be more but during the period of time you become older the ice creams You might take a couple of scoop but in a childhood you might take a so basically you're just trying to understand the pattern. what it helps you is, okay, if I buy two boxes of ice cream, this particular customer is a 20 year old, probably he might go, his parents are just sure. It tells me that he's an ice cream lover. So let me produce this, these offers for the age group between, 20 and 30. Right? So, there will be four algorithms that easily can help you with this. And one is the collaborative filtering. And the second one is the content based filtering. Where, the customers are already interacted with based on what they have, purchased. Okay, then comes to the neural network where I call the deep learning side of it trying to understand the pattern and predict his future preferences we all know. you know Say if It is going to be winter right for example What will be a pattern of products or anything that a person might buy during the winter? so to just give you an example, it would be very interesting, where a customer, he always buys some, some pills, say for, for cold pills, because you have the option, you have the tendency to get sick during the winter by cough or cold or anything. So whoever purchased that, purchased that in that particular store. This store opened a pharmacy. this retail store. so think about like, why would a retail store open a pharmacy? Because the patterns are like showing like that, right? you buy the, based on the product. one part of the department, you buy the grocery based on the department, he might have the tendency to, to buy, the cough or cold, medicines. so this all comes from the deep learning analysis, based on your, behavior, right? And again, examples like, if you go to, right now, even if you go to a retail store, previously, any retail store is, come into mind, say for Walmart, 10 or 20 years back when I, 10 years back, I was going to Walmart. It was just, groceries. Now, they literally expand it to the clothing division. Right now I can go and buy my, they expanded to a pharmacy inside it. Why would you buy, why, there is a pharmacy like Walgreens and other stores and why would Walmart get into the, but this is all comes from the Patterns that the customer have bought from the store. Okay, so it literally means like you bought something from the store You become sick. So rather than going to the pharmacy is right there in the retail store, right? Then the interesting part is now you can see that the groceries are there Yeah, the rest of the things have been accomplished now you can see The restaurant the small restaurants are inside the walmart. they think have you ever seen like you're going to a walmart and you know buying pizza or Subway, you know because people are tired. They know that they will be hungry by the time they pre shop so rather than just taking their car or just going out to subway somewhere Why don't we just put the subway that right there? see this is all comes from these kind of patterns four algorithms we have used. One is a deep learning and the last one would be the natural language processing. This analyzes the textual data, reviews the feedbacks, understand the customer's sentiments and preferences. So these are the vital thing that we have used, that I have personally used in the, in driving a personalized coupons. okay, moving on, moving on. Why are we doing all this? What is the benefits of this coupon personalization? It is pretty simple, right? As I just informed you, one is improves the engagement. most of the time, keeping the customer, engaged is so important. As I told you, just to give you an example, why would I have a restaurant in a retail, store? a fast food store in a retail store just because, like you are having customer engage to buy more things, right? Say if I, I was in, in a store, right? while printing the bill, it was, I bought some, $50 worth off, products and it just stays like you are worth to buy, a pizza for free, a slice of pizza at the food counter, but free. I was literally thinking of after the store, I have to go and buy food outside to go to a restaurant Or a fast food and i'm just I don't know how they read my mind. Like I was really hungry It was like 2 p. m in the night. So bye i'm just Okay, I love it, so it just gives me that personalized coupon, based on my purchase. you go and have your free pizza because you bought a 50 worth or a hundred dollar worth of product, from the retail. So this is wise, right? So have them customers engaged. The next one is, hyper AI driven, AI driven personalization can drive significant sales goals for the retail. trust me. I have just put it as 150. I have we have seen the retail store making a huge amount of profits, and they can Decide, the based on the customer like what all that they need to buy, you know What all they need to because have you ever seen? retail store dumping their products outside that has become less They know What that particular store or the people might buy they have understood the customers better say You know most of the time I have seen The grocery products like potatoes or anything. It's always You know or a pre diff time. It just goes where they just dump it in the yard But right now they know that okay, this particular store only needs this much of kilos or pounds of potatoes So they don't over buy so ready by reducing that they have They have raised the bar like in they have increased the sales and they have saved And the most important thing that I wanted to mention here is the customer retention personalized coupons has become really a built a loyalty, for those customers, say I use target and walmart most of the time They have their own and differences in pieces, but I use walmart the more because of that you know if some store is giving me that personalized coupon, I would love to just go back again and again, because, they know what I wanted and if I'm just getting a better price out of that, why not? and it just definitely increased my boost because based on my purchases, they're giving me discounts on other things. Say, for example, I told you about a free food that was, based on my purchases, I got a slice of pizza that is more. But for me, I, I like it. I was hungry. I know they understood my preferences and there was a coupon that was just popped in my screen. okay. Based on your purchase, you got a free, just go and have it, what else you wanted. Okay. Next thing, cost effective marketing. this is very important. the demand and supply, it's very cost effective. Rather than, giving a personalized coupon, pretty general, which has so many things, and I don't have time to just sweep all the coupon pages and see which of the coupons will fit for me or fit for me. Maybe there might be a coupon that might fit for me, but not for that time. Pretty simple. Okay, moving on. moving on to the next slide. Okay, definitely, I cannot just talk about without any examples. There are case studies, and, I have, I have made it as a very general case studies. Many of Amazon. Many of Netflix. I've just stayed at Sephora. That is a, that's a different, that's a big retail in U. S. So I will talk about that as well. So one is Amazon. Look at Amazon. why they have been very successful, right now is just because of, the recommendations that they, they give. based on your purchases, they have collected all the data over the years and giving you the right relevant product and the discount that really makes sense. And, I don't know. I have been using Amazon for the past 10 years, and I have, and I don't want to go for another one, even though the other people have given me more offer or the store like Walmart, but still the Amazon gives me a little value add than the other stores, especially in the past. On the coupon space. right now in U. S. there's a prime deal going on. I love Amazon. After this meeting, all day I'm in Amazon just because of these, because of the coupons. though I don't want it to buy because of the people offering me the best coupons, I just go and buy it. Damn. Okay, next thing. Netflix. Personalized content recommendation. Okay, I think we talked about a lot of algorithms, right? And, Netflix is the pioneer in using these, algorithms. Netflix, they were going down, and now suddenly they just picked up, their shares went so high just because of, they are using, definitely they are using these, they're giving the, they are using, the recommendations, you know based on the movies that you watch predictive analysis and the deep learning and the natural language processing they understand that okay This person likes to watch romantic movies or thriller movies, such as that's something just based on That particular Larry, how do I do it? you see it's pretty simple, right? I have a Netflix account, one account, but we all as a family shares, my wife wa watches Romantics. I watch as Thriller. My son watches cartoons. So based on that recommendation and personalization, you get into that particular, account. You see all these priorities. It just personalized for those particular, persons. And again, the last one is a Sephora, it's a virtual artist apps that enables customers to buy product virtually, imagine leading a higher customer engagement and sales with users trying 50 looks per session on average, I definitely you should use one of those it is it is great. Okay, moving on to next. Definitely, if there is a technology, there is definitely, there is, there, there is, there are high chances there you, you have challenges. Okay, there's nothing that, again. one of the biggest challenges what we have, seen, in this industry is whenever you're collecting a large amount of data, because, you're trying to understand the customer's behavior. especially I told you, if customer is sharing an information saying that if I am diabetic or if I have some other disease, those customers should not be recommended for a more sugary products. It's just a basic understanding. So there, you're personally collecting those information. You're trying to understand okay, this customer. So why would I, have to go and tell the whole public of the world that I am diabetic or I have this problem? But, you're literally entering that data in that app, in the re, in the retail store app. That has to be preserved, right? So the, so encryptions or anything that retail store, the data privacy. Because once you hack, security and hacking, I see many people talking about security and cybersecurity. I talk, when I say data privacy, you know what that means is okay, algorithm biases, AI models are premature by us. Resulting in unread treatments across different democratic groups. Okay, this is so funny that, when I ask a question, to chat GPD at the same time, I ask, at the same question to GE Gemini, or, the Google's version of it. And I use prosperity of, There is another, there are so many, every, GPT gives me a different answer. Some of them, it just hurts me. Some of them I feel like, why would you just put, put the state by a statement rather that is not right, so you've got to be very careful in those using those algorithms. say for example, you, you cannot, say for example, you, if a kid is being in the store and a kid, a person of less than 18, who is a customer. Uh, you are no willing, you cannot recommend no matter what. You cannot recommend, an alcoholic beverages for that kid. though the kid might be searching, based on okay, or he might be doing that, but, uh. you're understanding that, okay, this kid is, he's less than, seven, eighteen years old, so I'm not going to recommend or give him promotions on an alcoholic. this is a basic, understanding. I'm just giving you a high, very high level understanding because I don't want to get into the bias things. I don't want to become, more religious, and just, just telling, what the bias means and, I don't want to get into it. past. Algorithmic biases are very common. when we are using, these algorithms, just make sure, we put a class that we understand. for example, recently, like my, my, one of my friend is a professor. pure vegan. and when he gets an offer for, for, meat, he gets so offended. Like, why am I getting these offers, to buy chicken? I, though I am, I'm a vegan, right? So I'm just giving a very high level bias statement that what that algorithm predicts, and, and just, It just gives the results on that. And next is the trust and transparency. customers trust before they put that information. So customers find overlay personality offers innovative, right? ensure the transparency, and giving user control their data is crucial. when you're collecting some information, just make sure that customer know that you are, you're collecting those, information. I think at least in U. S. there are many companies that have been questioned by, by the government, saying that, whether Facebook tracks are, where are you? what they do, right? So I don't, I'm not getting into more detail on this, but just in a very high level, How trans, we have, we, as an AI or whatever we are building for the customers, we got to be very transparent with them, with what customers, okay. Technical recommendments. Implementing AI is expensive. AI systems are pretty expensive. It requires significant inventory. So think about, I just told you about, at least I need a year or five years worth of data of what customer has been doing. Purchased in that particular store over the years collecting the data, cleaning off that data and make it a valuable, a truthful or, or a very, meaningful data to the, to the AI and ML is crucial. And it is, it's still, you need a lot of skills and, time infrastructure cost. yeah, these are some of the challenges that I have faced, when, when you're building machine learning algorithms. Okay. Going on to the next, trends and future, maybe, future, future trends. what are we trying to deal with it? Okay, so as I told you, this will be your future in the retail space where nowhere you will be getting a very broad, a newspaper or a very broad okay, these are the things that are in the promotional rather than it will be a personal. So AI is going to enable the real time adjustments to promotions. based on the immediate customer behavior. I gave you an example for the pizza, okay. Immediately I bought it for 50, 50. You just try to understand my personal behavior. I, they might know because they want me to have, have their lunch or dinner at this time. At that, at the retail, rather than just going to a restaurant or a, or, at a fast food place. So it is gonna be dynamic, right? why does DI dynamic? Because I bought something and I don't want a coupon for tomorrow to go, okay, go back to the retail store to have a pizza. Why would I use that? I don't want because I go to two weeks once you know if my coupon just arrived on that time While I am just going heading to the store. How valuable is that thing? Okay, so it is going to be real time offers Will be producing for customers deeper integration. So AI will expand beyond the promotional strategies Okay, like customer services Automation, Inventory Management, and Predictive Analysis for future customer needs. This is very important, right? So the AI is going to, engage, will be more, personalized. So say for example, if I am, if I am, going to Walmart, the app or the AI app is Walmart app, just gives me a, gives Tells me that, hey, if you're about to, hey, this is Sunday, I understand, if you're going to Walmart, this is the offers that we are delivering, just telling that, right? Because I don't want to just go to their website to understand what the promotions are going on, rather than AI just telling me, you go to the store, this will be just for today. if you're buying a laptop or you're buying a phone, there is an exciting promotion this Sunday. If you go, you're the one who will be luckiest to get that offer. How cool is that, right? Okay, and, A, augmented shopping. Really, retailers will be offered an even more interactive shopping experience through AI, like virtual assistants. Like AR product previews, right? most of the time, okay? you're always in a dilemma to buy something, right? You go, you pick a laptop. you always ask the question, whether this is the right price for me? whether this will be useful for me? what would be the return policy on this? what will, what are people talk about the reviews about this product? And you got to be just keep searching for all those, and just to come to a conclusion rather than think about you see a product Tip, tip. These are the results that you had, have, this many reviews have been given all are like 4. 5 and more. And, by geography, this is one of the best laptops that you can purchase from the store. Okay, this has the best RAM. This is the this is a jump it just pops up in my app or application about that product, it just really Makes me a touch point like I don't have to spend too much time whether to buy that or not Right and give me what are the advantages of and disadvantages of buying the buying that product. Here you go you know I don't have to call my friends or people who have purchased and just spend my time on a call or on a To understand about the product rather than it just boom pops up All right Conclusion, good, this interactive of AI personalization promotion strategies is transforming the retail industry a big time, offers businesses the ability to deliver high category and more relevant coupons, as I just mentioned this, right? And, I'm just not going to technically, I'm not going to read about that, but just in a high level, 35 percent raise in coupon redemination, think about like how many people are redeeming your coupons nowadays, you know it I don't think you give Any attention to coupons or anything or for if there was a sales guy just coming to near to me I'm just going to walk away from him rather than just Standing with him and just talking about the sales promotions, right? this is going to help in 35 percent rise in using the coupons And, we have found that there's 22 percent of the customers you will just retain with that particular store, right? However, while the potential AI impersonalization is immense, businessmen also address the key challenges that I have mentioned on data privacy, compliances, there should be regular guardrails for all of this, all of this, technology behind, that works behind the AI. so somebody is not misusing that rather than just, it is just going to help the real man, help the man, mankind. some companies that are developing the artificial intelligence, models, machine learning models should be more cautious to be, to understand, okay, this is sensitive data, I'm not going to screw up with this, just add, I'm going to add some extra layer of protection on top of it. Got it. I think that's all about, the today's presentation. I hope you all might have enjoyed this. And, if there are any questions, feel free. My contact information are, are there with the conference, web page. and, it was a pleasure. to give me to, to talk to you about this and, connecting and meeting you or everyone on this. thank you. Bye.