Arpan Biswa: what AI changes—and what it does not

If you look at the online case of users right currently, there are 10 CR users out of one portion and divide into different cohorts which will improve the overall efficiencies. And the challenging part over here, it has to look on real-time. Before we implement the AI, I think we have to take care of natural super-series before looking at artificial intelligence. What data do you check to understand which cohorts you want to put the customer in? Until and there is a lot of real-time learning, it becomes very challenging. I wanted to launch a collection with Urfi Javed and Najee as episode Frank. Arpan, I'm going to come to you. Like food, fashion is for millions. And has various cohorts and I think you guys know better how to segment that.

But how do you tailor messaging using our today in the world of AI, messaging based on customer emotions, understanding the emotion, where they are, what they want to buy? That would be great. A quick introduction. Arpan, head marketing at Reliance, Ajoo. Taking the discussion forward, I think cohorting becomes extremely imperative for us primarily because if you look at the online base of users currently. There are 10 CR users out of 140 CR of India's population and we have recently touched 146 CR, the number one, the most populous country in the world. Put it around. Only 10 CR of those guys are currently online coaches in some fashion. The rest of the guys are in and out. They are not completely moved from offline to online.

Now, if you have a certain base in mind and Ajoo being one of the newer players, for someone like Myntra or some of the existing, like the Flipkart and Amazon, which have already reached out to this base, it's more important for them to dissect and bisect and divide into different cohorts which will improve the overall efficiencies. And the challenging part over here, it has to be done real-time, right? Or else you can't go and introspect data, go back to them and show them certain creative and certain tech property and know what you're talking about. Now, keeping that in mind, the biggest challenge which we see right now. Earlier, customers used to be divided as per demographic and psychographic. Right now, the division is as per their mood.

The same person who is buying a super drive, which is one of the premium brands on our platform, is also the person who comes back and is driving a 300-dus expenditure. Now, how do you understand basis, what data do you check to understand what you want to put the customer in? Until and then there's a lot of learning, real-time learning, it becomes very challenging. AI or, I always say that before we implement AI, I think we have to take care of natural super direct, before looking at artificial intelligence. Now, these are the basic things we need to take care of even to understand the user, where he's coming from, what is his mood at that very point in time, what is his behavior before coming to our platform,

what is the awareness level for the brand, given that how he'll interact versus any of the established brand versus a new brand, he'll interact very different for the same product. There's so many different factors which has to go in. The biggest challenge which I see for that is everything coming together under the same platform, wherein the tool or the platform learns from all these different variables and come up with a model which predicts correctly. That's insightful. About email and delivering that email, I think what's all they, at the specific time, somebody reads it. So brands are today using AI powered predictive analytics to actually optimize even email subject lines,

sheer timing and using machine learning to personalize content on individual customer preferences. Old school channel, still a very effective channel. So for a business like you, I'm assuming you use actively email. How do you see this changing? I think not just email, let's look at all the different CRM platforms for safe push notifications, emails, SMS, WhatsApp. We have, what we do currently is look at the data of the last many years, when you are sending a push notification at 6, when they commute, is the highest percentage of CPR or the conversion is the highest for that. What kind of messaging works well? So once you feed all the information in the system and have the right cohorts of users,

you have more than 100 cohorts of users based on their previous journeys, previous purchases, what kind of brand, what kind of APB they are purchasing, you can target them very well. And it has also helped us improve the overall efficiency and secure of the world platform. I just wanted to add, after this, I just wanted to add something to Jitain. Yeah, yeah, please. Recently, we had done a campaign with Urfi Javed. I know at the moment someone hears what Urfi Javed. Bipolar view, right? Yeah, Bipolar view. But I think as a marketer, we should explore and as an enthusiast for fashion and from the fashion industry, we should be open and inclusive about everyone's choices. Right? So we did an April Fool campaign with her.

I wanted to launch a collection with Urfi Javed and RGO as an April Fool prank. I don't know if any of you came across it by any chance. What I was looking for, I reached out to copy head and I asked him to create something really fun, which is positioning say a duct tape or a basket or any of the items she uses, right? to cover herself as a fashion actress in Shashi Tharoor's English. As a limerick or a sonnet, right? Which is slightly satirical in their tone because it's a April Fool prank. So when I briefed this guy, firstly that guy has to go back, understand the entire context. What is Shashi Tharoor English? Learn that. What is the limerick? What is the sonnet? How are you supposed to write that?

Instead of that, when I prompted the same thing on ChatGPT, it has all the worldly wisdom. It doesn't take time to learn. It's only learned. And within seconds, and I had to go live in the next few hours, we got delayed for whatever reasons. It gives me such lines which work so beautifully well. When someone read it, they fell in love with the lines. It's actually in Shashi Tharoor's terms he has coined, right? It's already in that. So I think in few very specific functions like copy, I think the role of ChatGPT and other such tools is immense. Because you don't have to go back and learn or brief someone. It's already a part of their inherent native wisdom and can just start delivering from that very second.

We spoke about AI and ML, segmentation, messaging, and cohorting and understanding emotions. But how do you see AI powered virtual assistants actually providing personalized shopping experience towards the end of the journey of a purchase journey? I think this is one of the things that's in work for us right now. So I think the belief has always been when the customer buys from you, that's where the relationship with the customer ends. Not the case anymore. In this connected world, that's where it begins. And given that there's so much of competition, you have to create brand love through all these new products. And be right next to them as a friend who can help them with their choices. So we are working on something very similar right now.

This is our previous buying behavior. A lot of times they also upload. They can upload photos, image searches and photos based on which we try packing data. What are the choices and stuff. And recommend things to them. This is their fixed, if they're buying full cutouts or they're buying more red colored products. Accordingly, push that to them. That's something in works right now. Which we are working on. Great.