Hello, everyone. welcome to the con 42 JavaScript 2024 conference. And my name is Ananda Kumar Abellu. And I am a cloud consultant and an AI ML consultant at a company named Sara System, who's a premium cloud partner with Google Cloud. We closely work with the Google Cloud team, and I'm one of the consultant work for the company. So today's topic is what I'm going to talk about is The impact of machine learning on promotional strategies, which is like a study of, AI powered coupon personalization. so I'm quickly, let's get into the topic. let me start with the, with what I'm going to talk about today. So one is the introduction to AI powered coupon personalization. So when I say coupons, it is just that, You know when you go to any store, they give you some special discounts for a product or anything that you buy comes with an additional discounts so It can be either you're going to a particular store or it can be from an e commerce site or anything that you buy, the discount, that you get, for that particular product. So AI powered coupon is how the AI decides, how much discounts or anything that can be given, To a personalized person. Now the coupon is not for everyone, you know the coupon when I talk about is just all in general that we have but how that particular person or that individual who's going to buy would get a discount for that particular product or whatever he is going to purchase. the next thing is what is the fundamental about the AI powered personalization? Why AI is powering that, how or how it is going to identify that particular person and only for that particular person AI is suggesting for, a coupon or a discount. So how these, personalized coupons are and offers are created. the fourth one would be what is the benefit? Why are we doing, this AI powered coupon personalizations? And the real world case studies, there are some companies still follow that and they have been very successful doing this, coupon personalizations and i'm going to talk about that and the next thing is what are the challenges and ethical considerations that while doing this doing this and what are the challenges that we come up, from a security, from a data ethicalness, from data governance, I'm going to talk about some of the challenges, of that, and some key performance indicators when you are doing something, it's obviously you have to calculate the cost. KPIs and how we measure that this particular ai, coupon, personalized is working for that particular company and as well as for the customer who is buying that. And that, the, and, and also where are we heading, what is the future direction for the ai, AI on these promotional strategies? And I'm gonna, conclude with some of the points, why I have been, Consulting or, consulting this, coupon personalization for, big enterprise companies as well. thank you. And, let's, quickly get into the, contents. So first thing, introduction to the AI personal coupons. So let me tell you, gimme a small stories. You remember, like I go to a store example, like a Walmart or a Kroger or, the nearby federal market, anywhere. when you enter the store, there is always a bit of paper. Where do you see there are, when you pick out, you pick two pineapples, that there will be a third pineapple for free. you will have a very generalized way on the new, like a newspaper. if you buy one kg's of oranges. there will be another, kg of oranges for the half the price. and some, sometimes like you buy two shirts, you get the third shirt for 50 percent discount. you might always think what is the logical logic behind that. Yeah. Yeah. Logic Behind it is obviously they wanted you to buy more, but at that time, most of the time I have seen that you really, I don't have the intention to buy a shirt or a pineapple on that particular day, but the offer is there, right? But I really don't want to buy, why would I buy? Because I just bought, a t shirt. a shirt or a pineapple from a day before and when I go the offer will be there for the next day and it's always a disappointment that I have encountered. I'm sure that you might also seen that same scenario. Most of us. Let's do that. Oh my God. I just bought this, but today the offer is on, on, on, on these today. it is always the advertisement is, is being wasted. Some of the time people buy it, take user, Oh, I think of doing it, right now there is an offer. Let me go and grab that. But it is not the case for all. and in a retail. e commerce site or a retail store. This is how the traditional, coupons are, working. so the, there is no way where whether this coupons that been advertised, or sometimes you get a mail as well saying Hey, you go to the store, this is how, you buy this, there is a discount. But at that particular time, at a certain point of time, you don't need that product or you might have that product already in store. But when you want to go and buy, it's always the discount is not available at the time that you decide to buy. I hope that makes sense. So what, what we are trying to do is we are introducing the AI in this coupon personalization. And, I'm trying to see how this is, this helps in this, personalizations. How, how, better this whole strategy of giving discounts can work. Okay. Obviously, AI is being, AI and machine learning is top in the market. it is being used, widely used in every industry. even I have created myself an AI bot, which it, which I talk to him. You can ask me who me, for every person, AI is big. Right now being a part of our, ecosystem right now, in our daily life. this is again a revolution like how we cannot avoid phones right now. So AI is the next move where, we are going to use the phone for, like phones, AI will be a part of our, everyday life. as I explained, the role of AI in everywhere. The second, the personalization matters. I already told you the traditional way of marketing methods, why this matters because I, I wanted to, I really wanted a discount, but at that time the discount is not available. Rather, whenever I wanted, the discount is not available. Whenever I, whenever I don't wanted it, the discount is available. As you see that this is an e commerce market, it is rapidly growing. it will reach definitely by end of this year is a 6 trillion market by 2024. And how these AI can get in, AI powered coupon strategies can really make a difference. differentiation in this e commerce and the retail stores, in this current, economy. So that is what this is all about. So the key concept, AI personalization uses customer data. Pretty simple. I'm getting the data from, a timely ma matter from, from a person. Say for example, I'm using my credit card, I'm getting, I'm going to the same store every time. So it is just like trying to understand the purchase history of that person and try to give, the personalization. so for example. you remember that is always I don't drink alcohol. I don't buy beers, or I don't buy any alcohol stuff. So when I go, there is a coupon for beer. Like you buy one beer, there was another set of beer or alcohol is there. free or there is a discounted price. I don't want that because I am a non alcoholic person. Why do I need that? So try to understand these kind of criterias based on. for example, I am an apple lover. I always buy apple. It will be better, from, the, from January till October by this time. I have almost like bought, Seven or seven, eight times apples, like almost every month I'm buying apples. based on particular this history Okay, this person likes apples a lot. So he's been constantly buying apples So I want that person to have an offer only, only for him So he buys two pounds of apple. The third pound is really free for him. So You understand where I'm coming from. It is just personalizing some, for that person for that particular product. Say it can be biscuits. So based on his, based on his purchase history, this is one way. And I'm just going to talk about many ways how this personalization, works in the coming slides. okay, quickly, next one. What is the fundamental? What is AI personalization as I just, told you, it involves the, some of the machine learning algorithms to customize the market content and promote offer based on that individual's performance. Say what I conclude is based on my health history. One is health history, my purchase history, my age, my, my. what else? I can say my, based on my search history in the, in the website, in the internet, based on my family, members, it can be of any, any kind of scenarios. The best is the way, is how much or how we can collect data from that particular persons and know that particular person, through mul various sources. So the key components are how the data is collected through these systems. as I told you from different ways, from different sources, these informations are collected. Then based on that, the patterns are being analyzed. as I given you an example of apples. So if I buy apples every month, based on my apple purchases, the coupons, The base, a coupon is being generated based on my history. So based on my history, I am a, since I'm not an alcoholic, I don't buy alcoholic beverages, I, we don't need that particular coupon for that particular person. Rather this can be given to a another person just for who's more interested based on that person. His purchase history, for example. So here, there are some of the tools like how the predictive analytic tools. so say for the last five years, for every October I am buying something that only I buy in October, so it just reminds these, it, it is just a prediction. So it can, the purchase history can deliver, tell, okay, this person, for the next October he's going to purchase this, Halloween costume. Okay, or even the during the Christmas time or maybe my in October I was born So he's going to buy a birthday cake every October. So why don't we just come give a five dollar or ten dollar? offer on that particular cake that he's going to buy So, you know what, it's just, it just gives the prediction. based on your information that you have already given, this kind of like just predicts that, predicts, the algorithm predicts what it would be the preference for that particular person, to issue a coupon or, So as I just mentioned, some of them are the purchase history, browsing behaviors, demographic mean, what location he's in, he's into. I live in a very cold location. it is right now, it is 45, degree Fahrenheit. So based on that. so it just predicts me, okay, this person might buy a woolen cloth or something that, that, that just personalized for that particular, person and contextual data, consider the time of the day. the device used or the location, say if I'm just traveling towards, some couple of, blocks where there is Walmart, it just gives me a suggestion, hey, there is a restaurant inside the Walmart that gives you an offer, if you go and eat there, obviously if you buy something, in there, there is a coupon, that is offered, you get a free snack, in a subway that is inside Walmart, something like that. and also it just shows how data flow through the AI system and created those personal, just a feeling that, say if you go to a store and, there is already, you wanted to buy something and there is a coupon that is available for that particular day, it feels so good. Okay, it just makes me feel that, okay, I'm getting from a discounted price that is not for others. It's just personalized for me. it? Okay, let's quickly get into, how this has been done. so here, I think as I already mentioned in the previous slide as well, customer profiling is more important, than everything. As I told you, the age, demography, his behavior, Purchasing. so how, what he would, buy, what he might be, his interest that he might buy. say if, he loves, apple, there might be a chance he might love a beer or, piers or grapes. There might be another fruit that he might be interested. So based on that, if you keep on buying apples, why don't, he may have a tendency to buy oranges as well. So it just gives us, throws a suggestion on, that, and this all comes from, the customer profiling. Then comes the predictive modeling. AI analyzes the history of data and, as I said, forecast, it forecasts which coupon offer will apply for each engagement. so say if I, if it is a weekend, the discounts might be different. And if I'm going on a weekday, then it is a different. different. so this, these kind of things just goes in, in, it, it called as a predictive, modeling, it's just based on that particular date, time and, and the month. the real time personalization, it is a dynamic generation based on customers current behavior, such as I mentioned, is about browsing history, product category that you might, buy, and ensuring the timely relevance. That is more important. And the word timely relevance is, I don't, buy. Buy, a log of wood in summer, rather, a log of wood keys always bought during the winter, when I'm go for camping. so this is the relevance and the relevant discounts that can be offered during, that the, on that, on the time, based on the time. So next case study, one of the. Best and the pioneer in using the personalized coupon. Coupon coupons are, Amazon. Amazon has been a great, and they have literally boosted their sales by 31st through personalized suggestions. And you can al always use, if you're using an Amazon app, then you will re realize that, there are some suggestions which just, really makes you to go and search for those, search for it. you might be just. thinking about it or you might have just searched somewhere else and automatically just gives you that particular suggestions on that particular product. it's unbelievable. Like it's all like you might be thinking, and that, there is a suggestion based on the app would just show that particular product. And, and if you're just wondering how I was just thinking, how come this is there? And, I, this is all done by, definitely by the AI. Okay. why are we doing this? What is, what is my benefit in doing, AI powered coupon personalization? one is, your, boost. Boost is the, engagement, you are giving, more, more offers resonate that more to that particular customer leading in increase interactions and satisfaction. pretty much close to the customer. Basically, you're trying to understand what customer would want it. if you, understand the sales, you, you should know your customer. You cannot sell something that customer won't buy. You have to Put in your shoes when you sell something, it's a so trying to understand the personage and it also drives the sales growth so something that you are keep on buying and which is not sellable. You don't buy that You only wanted to have something in your store that can sell easily or we can sell in that particular in that season or particular time. So it just gives you, a lift in, in, in the, in your sales, in your sales. So it persuasively sees 30 percent and, another important thing, as I said in the beginning, those newspapers that you, when you enter, those are all like, it's useless, for me, I don't grab it. Or if somebody is. Just pointing me, I don't go and buy that, because I feel always that, it's not gonna help me. those coupons are not gonna help me anyways because, always when I get it, I have seen that, I wanted something that, that those products will not be an offer or, there will be no coupons for that. by this way, 30 retail stores on the e commerce site have seen 35, 37 percent is increased in the redem, redemination rates. So the offers, whatever that been offered, they have seen that it has been redeemed, redeemed 37 percent increase. So increased retention. Why would I just go to another store which just gives me a general coupon rather than I know that this is the store which gives me more than what I wanted? I always try to pick that store, right? That's called personalized experience. it creates a stronger emotional connection between the customer and the brand and increase the loyalty. I don't, I would just buy a same kind of product, every time if there is a discount available. why would I just choose another brand? It's quite obvious. And cost efficient. always like you are predicting that customer is going to buy this particular product and you can, you can, you can be very, the store can be cost effective as well. And it can also, be very efficient for the customer that he gets, what he wants what that he wanted to buy for a better price rather than, rather than a, like a standard price, from another store. he knows that he's going to get discounts when he, whenever he enters that particular store. better allocation of marketing resources, reducing promotional spending waste, and the best part is the ROI improvements. And the study shows that AI personalization would double the return of investment, obviously, in the marketing campaign. So whatever you are, spending on for marketing and, I, I've never seen that it is going to really, be in vain. And the last one is the visualization. The bar graph, comparing traditional with the AI is a key performance matrix, like conversations, rates, retention, always there is a high in, in everything that you do. Okay, moving on to the, next. Obviously, I told you that I will be talking about the real case studies as well. so, The first and my favorite case study is always the Netflix, because Netflix has been a very great in, in doing this for a long time. They've been using, before, AI and, ML has been very popular in the industry. have you ever seen, Netflix been recommending, a pattern of movies that you always say if you're, if you always like thriller movies, which I love, it always recommends me like whenever I turn on, okay, based on your, thriller movie suggestion, you might also might like these movies, which just makes me to either buy or use that particular, Netflix, channel all the time rather than me going to another, in a real case study. I think, Netflix have become an estimated 1 million annual savings, because of just by, giving, these personalized contents to, to, to customer. And the second, thing is always, Amazon, I think, I already spoke about Amazon. I have never seen anyone, who, I'm being amazed whenever I use that Amazon. I always use Amazon. I think I use two apps. One is Walmart and Amazon. I can definitely just make out a difference between Amazon and Walmart because the recommendations and the discount that, Amazon offers is fantastic. Fabulous, because they do a lot of collaboration filtering and deep learning algorithms and generate and generates 35 percent of the total sale just by giving those personalized suggestions. And always, whenever I go, there is a coupon available for me if I'm just going to buy something. I don't know how, but, just wondering, I'm sure that the algorithm is the AI and ML algorithms or the background of that. And some of the other examples are, I'm not going to read everything, Sephora virtual assistant. It just gives you, it's an app where, it just gives the leads like 50 lookup average sessions, increasing purchase intent. It's an online engagement and driving sales, both online and in store. And there was a takeaways, another company, which is an e commerce and, to digital content streaming and cosmetics, cosmetic retail. It's just another example. These are some of the examples and real time case studies. Okay, moving on. challenges. whenever know, whenever there is something, even in IT, there is a huge amount of data is being collected. and this is all done in an algorithm. you might, as based on our conversation, there is a, there are There is so much of data churning, data filtering, data being analyzed to come up with a personalized coupon. And whenever we deal with data, there is always challenges and always data can be manipulated. Data can be done anything, it can be used against you as well. So just make sure we are using these data in a very useful way for the benefit of someone. we just have, to have a few guardrails and number one would be the data privacy. The increasing, resilience on, professional data, data rises concerns about security because we are not just going to, randomly just, understand a person because we are collecting some personal data as well, based on, like age, his, spending history, uh, you, because you need to know the person, right? some of the health, health information. Say for example, if a person is diabetic, okay, you're not going to give, a coupon to buy more, sugar or, or a very sweet carving, products, right? So obviously that, that, that guy is not going to, based on his. based on his health, he's going to decide his, future. And, and you don't want somebody to know, that you have some health issues. Obviously it is just, it's just me. talking about, you, there are some security and regulatory compliances that we want every company to, to have. to follow. those are CCPA and G-D-R-P-R and business must be transparent about the data use and you're just gonna, give the information to these companies based on, and, and, and we also wanted to know that this companies is using my data, to generate the coupon. make sure everything is transparent. algorithm bias. The AI models can inadvertently predicate bias based on the demographic leading to unequal treatment. Okay. as I just mentioned, so regular audit should be made, you don't know that I literally touch based on something like, a person from a cold region, what he wanted from a heart region or from a race, or he can be from a different, demographic, location based on that. you should try to understand, say if, for example, right now I'm living in a colder region and I just moved. My, my, my, I just moved means I've just moved everything to another place. So it should understand and should not be biased based on, okay, this guy, he's not just going to buy, this, maybe I've been buying this for all my day, life. So still the coupon personalization to work. So each has to be a constant check. every time I, we buy something, just make sure, that. the algorithm is reading based on where the customer is buying, which of the store he's getting it. So just have a track of the customer. next one is balancing personalization and intrusiveness. over personalization may feel Allow customers to control their data and provide transparency on how it is used. again. Sometimes you are over personalized, you may feel very bad, every time, over personalization, it can be based on the history rather than giving suggestions that you may not, say if you're buying apple, I don't like grapes, why would I need based on your fruit buying? History you cannot suggest some Fruit that might be harmful that for that particular Person, so just don't over personalize Personalization is also a bad so make sure that you have a customer have the control say there should be something Invariant an option where it could say, don't suggest me any of these fruits, something like, or don't suggest me any of these products, though I have bought them in the past for someone, but I'm not really interested, so there should be some guardrails on that as well, show me these, or don't show these products. products, which are, not relevant. for example, the product should know that, I'm a man. I might've bought something for a woman or my wife, but it should not just give me a suggestion to buy a woman's product all the time. I'm sure like you, you would understand what is over personalization means. I'm just, let's get into that next topic. Okay. technical challenges, required advanced data infrastructure. skills, AI talents, this is always a biggest challenge. Though there are a lot of, AI enthusiasts in there, but the investments for doing the data analyzation, data, data warehousing, data cleaning, this is all a biggest challenge. in the recent days right now, until there is, very high much demand in the skill set of AI talents, the person who is deciding on working on that, who really have to have that knowledge on what he's trying to do. there's definitely a significant barrier for small companies to come up with these, with these technical challenges because the infrastructure that is, that's been built for these are. are a lot and is expensive. so highlights, solution approaches like regular algorithms, audit, data analyzation and customer consent management tools should be used. the solution approach has to be, audited at least six or three months every or at least a year. so these are some of the ethical, and challenges. for doing this AI coupon personalization, let's move on to the next. whenever we do something, it's always that you have to measure that, right? say, I would say the progress. So this year I have accomplished that. What is for the next year? It's I think we are all most professionals working in those companies. we are always tied up with the kpis End of the year. It's always a question your manager asks you what kpis are how it is not just you it's for any product or Anything. How have you? done from the past year to this year. these are the key performance indicators. One is customer engagement. Have you really got success on this? how many click rates that the customer, click through rates, the time spent on the page, the interaction in that, and the, gauge interest that customer had, this, These are some of the KPIs that we need to measure, while introducing this, coupon personalization for any e-commerce or site. The next is, conversation rates. measure the percentage of personalized offers. you just have a track on, recipients who make this purchase comparing with the non-personalized, campaigning. That's a good point, see that, How many people that have really used this coupon, if they have not used it, try to educate the customer. Hey, you have some, something like this in your app, or when you're buying, you should, because it's based on your history, this offer is available. Maybe they might have not seen your email, maybe they are not, but I've used your app. So educate the customer. So have that conversation, rate And try to analyze that particular, particular subject on, on whether this offers that are being really redeemed by the customer. Okay. Redeem ration rate. That's it. so how many recipients do you use as the postional equipment indicating the relevance and appealing, appeal of the offer? So are these real, really relevant? some of the times they might show you something which is not even relevant. You just go back and check your data is accurate or is it something that, that is not correct. that is something that, you need to have a valid, validation check every six or three months here just based on whether it is working. customer lifetime value, so whatever we do, unless otherwise customer is satisfied There is no, no value on that, right? So you have a metrics like a KPI where you assess the impact on a long term revenue generation. From a particular customer, how much the revenue has been generated? I don't think many of the, many of them know, or any retail store how many times I use this go to Walmart or a particular store to buy this. And how much, I have generated for, for the Walmart, because I go to different Walmarts every time. I think, there should be a generalized way to monitor that. Okay. Then, KP analysis. Explain how to send benchmark. don't be shy to create new strategies. new KPIs, whenever you see something, you set your, and start, measuring those indications against those traditional coupon and distribution methods. Okay, talking about next, what is the future? the future is pretty simple. I know the AI is advancing into rapid growth. this industry has been rocking and it is going at a speed. I would say it's going at the speed of light. So there are so many improvements. Every day you re, you take, you read a tech newspaper. There was something or the other, in the, in, in any corner of this world. They've been talking about AI and ML and new innovations are being reported every day. So based on that, we need a future directions where we got, granular targeting. A advancements will enable more precision customer segments and hyper-personalization. so we are just getting more details and how I can offer some discounts in a very minute manner. That is, gang, granular targeting, integration with a, AR and vr. that's the future, right? AR and VR is so exciting. It's a broader chapter again. so I'm just going to put my glasses and just say, Hey, I'm just looking for some apples now. If I'm just thinking in my mind, it's almost like just done. It just gives you the coupon. Just all the purchases will be done. And next any moment my apples will be arriving at my door. So virtually try on and argument, the rear end. argument to reality shopping, you don't have to go physically inside the store, but regularly just put up with your AR and just do the shopping. And by the time you see something on the store in your AR, in your, through the glasses, it's almost like it just gives you that, particular, okay, there is an offer rather than you don't have to do an app search or you don't have to, go to the website to get a coupon. The coupon is already shown in your eyes. how fascinating and why is activated commerce? Okay, that's another, common and everywhere, I ordered my pizza through my, voice activated, devices at home. if I wanted, I just tell my Google home to buy something. So how about a Google home? how about checking, whether you have a coupon or something, in, through your personal assistant, or your voice assistant, speaker that can just, Alexa, just give me, I wanted to buy something, do you have a discount here? Yeah, it just throws you an offer, Pretty cool. Yeah. So ethical AI practices, increase focus on AI transparency and fairness, expect evolving revolution framework, and govern AI use in the marketing. Innovation trends. Discuss how AI can adapt a new data source like IoT, wearable technologies, and provide even more Contact ab, their pro, promotions that is more important, right? that's, these are all the, the future, future, this is all the recommendation, like innovation trends, right? which I would really, which is something that it's very creative, pro I call this creative promotions. moving on to next. We have come to the end of this presentation. as I told you, I think my message is pretty clear on what I have been. This AI coupon personalization, you would see in even a very small store, in any given point, in any time. You go to a personalization, small shop and you wanted to pick something, these personalized coupons will be everywhere. you can get into an e commerce site or, or a gas station. You wanted to buy something, any small, tiny shops, to retain the customers. These will, these, AI coupon personalizations will be pretty common. in, in the recent, in the recent years coming, in the coming years, all these, traditional newspaper, coupons will just be gone and, you will know from just looking, you get information, through your app or it can be from any of your, devices that there is a coupon available on what, where you go, by the time you go to the store, every time you enter the store, there is something that You know, you're thinking to buy something. There is a discount available for them. It's simple. is as simple as that. it can be electronics, it can be pa paper, it can be, vegetables. It can be, a meat, or you want to buy some. Food, something like, I don't know whether you, I have, recently I went to a store where I wanted to purchase something. I was, though, there, there was a coupon. They immediately just throws me another coupon saying that, Hey, are you having a food? I said, when I click okay. And he said, you just throw, Hey, there is a burger that is available free for you based on your purchase. How cool is that? and somewhere that is linked to the Amazon, store as well. so again, advanced machine learning algorithms, business can analyze vast amount of customer data to leverage high, highly targeted. As I, this is all, conclusion is just, I'm just talking about, whole presentation in couple of lines. So this is how, a personalized coupon, remediation is done. And the benefits are very clear. the business grows. it really, get, gets you what you wanted. build, it builds, a very tight, close relationship with that store. it just makes the customer happy. a happy customer, happy buying. it goes for the store. It just keeps the customer for a long term. and it also help. it also helps the person every time that you buy something, there is some other, it just values the money as well for both of the customer and the end users. the rest of the things, it is pretty straightforward on what I do. I am not going very technical on how the, how, what are the algorithms is all being used. But in a very general pay, high level, these are the, these are some of the, this is going to be, coupon personalization will be pretty common. I'm sure like everyone in every industry might like to, use it. it can be, it can be, if you want, can be to, if you right now decide to go and watch a movie, think about if there is a discount or a coupon available for that movie. How cool is that? Or if you wanted to go and watch a game right now, interest and you don't know whether there will be a coupon available, if there is a coupon available. just by checking, just by, based on your history and based on your, relevant, how cool is that? So that cool is what this AI. Personalization, personalization is, and I just wanted to thank everyone for this opportunity to talk to you and just through my insight and vision about the impact in, impact of machine learning in, in coupon personalization. And thank you and you have a wonderful day. See you soon. Thank you. Bye.