Ashish Tiwari: what AI changes—and what it does not

You can give me $1 trillion and I can give it out in less than an hour. The magic lies in figuring out people who will give it back. That if I give you money, will you give it back? That's where the key lies. The entire play of AI here is to be able to understand the triangulation of consumer need, which is unstated at this point in time. The moment it becomes stated, then he's needy and if you need a loan, you won't get it. You only get money when you have a latent need and not stated. So he has a latent need. That's the first data point. The second is he has a capability to pay it back. The third data point is very simple that he has a track record of paying back. And I process almost 20,000 loans every day.

That's the power of AI whereas you can churn a huge amount of data in very segmented way. the second is the future gazing session. Our panelists are going to probably to the next session. the next one. You need to understand my business. I am in a business whereas supply is a constraint, demand is not. So, I give out money. I'm into credit. Am I right? You can give me one trillion dollars and I can give it out in less than an hour. So, while for external word we say I'm in a business of credit lending, the real business is collections. So, actually my job profile is to be able to figure out that if I give you money will you give it back. So, here's the key lies.

The entire play of AI here is to be able to understand the triangulation of consumer need which is unstated at this point in time. The moment it becomes stated then he is needy and if you need a loan you won't get it. Let's be honest. The moment you need the money you won't get it. So, he has a latent need that's the first data point. And the third data point is very simple that he has a track record of paying back or he will take it in next whatever X number of days because of at the end of the day my basic my alpha as we call it. Technical term basically are typically average loan size is less than a lakh. I serve to NTC and new to credit and thin credit line which means people who won't get it from banks and things like that.

The AI here works beautifully because of if it was to do manually I would be probably deploying a force of a thousand people to do per customer. You have to create these blocks by the hour. The campaign does not run. Think about it. I do. I get almost about 35,000 applications a day and I process almost 20,000 loans every day. Wow. So, it's easy. Now, let me make it a little more realistic. Most of the loans, these are two different types. One is a cash one is a phone loan. Typically, people buy a phone between 6 PM to 9 PM. 80% of the load of those 12,000 applications is between 6 to 9 PM. And only on two data points you give you a Aadhaar, you give you a pen and I give you a loan. Sachin was talking about cash loan. He is the distributor.

He goes, the bank he was talking about is here. So, he only takes money from me. Okay, I was telling. He only takes money. I carry the load of the things. So, that's the power of AI whereas you can churn a huge amount of data in very segmented way. It's not only data, it's about the conversations which you have to have with the customer in the meantime. You can't keep a customer waiting on a web page or on an app page for three hours to get a loan confirmation. It's less than, so my journey start to end is exact seven minutes. You can get a 5.5 million loan or 5 lakh rupees loan in less than seven minutes. Only basis to documents. That's the power of machine learning and AI and data if you use it together.

On email, the entire journey, the reason I told you these cohorts is because if you understand that, now pull these cohorts out and put it into email. Whereas, if I need to send say him an email, I need to play this triangulation and then the email goes. So, from the timing of email to the offer I am sending out to a customer because if I send you an email of 2 lakh rupees and if I only approve 20,000 you are anyways, my effort is wasted, my cost is gone. I have to remain between the bracket of less than 10% variation. What I quoted and what I am going to give to you has to be in that bracket. And that's where email works very well. If it is backed by the power of data and AI at the same point in time.

And it's not only you talked about email, I am taking email, SMS, WhatsApp because of these are three methods. Today, if we think about consumer connecting at one medium, I think we will be fooling ourselves. We live in a multi communication channel life. I will see something on email. I will click on that and by the time I will get a call, it gets disconnected. I need to immediately follow back on our WhatsApp. Your journey is broken. If I don't do that, I will send an email. His email is already open and we saw that. His email is already open. The journey is broken because of you or she got a call. The guy has forgotten. You have to have a prompt at that point in time. Download my app. The app is downloaded but data connection is lost.

Immediately the bot has to dial down. So this is the entire journey. Three tips very clearly. Figure out your business case. Marketers, we all get very attracted to all shiny objects. Don't mislead yourself. I like what you said at that point in time. Tiptoe into this water. It's not that we have been building. HomeCredit is a pioneer onto doing these NTC loans. We have been building over the last 10 years and plus 20 years almost now. Do it. Pick up the smallest business case which is the simplest like you said. For example, we wanted to go in conversational way. We did not go to consumer. We have built an internal bot for employees. Whereas employees can ask thousand different kind of a questions. The bot is for agents, not for the consumer.

The voice bot certainly got trained like for more than an year internally before it is deployed and we use a prompt. You are talking to a computer. If needed press one to talk to a human. Cut the load but it helps. So first, pick up a business case that's simplest, the smallest, deploy there. The first thing. On email also you can use today all these conversational emails. Tools are available. Use them. And if you are not, somebody else is already doing. Second, use multi-channel mode of communication. Don't rely on one channel. It's not going to yield results. That's where AI is very powerful and works beautifully. Third, create your segmentations of consumers using AI. We all fool ourselves that every individual is one person.

And we keep talking about n equal to one. I am not the same guy when I am on LinkedIn versus I am on Tinder. Don't judge me. See, you judge me. If snail can get married, I am nobody to judge anybody. Great. Hello everyone. Ravindran Reddy or Ravi whatever comforts people addressing me. I represent brand Hindwear. Take care of the marketing for that. Many of the conversations in this morning. One of the biggest tasks for most of the marketeer sitting here is acquisitions. And most of the things we have been addressing post-acquisition stories. How to address the acquired consumer than acquisition. The biggest game for us because we are so-called magicians or so-called illusion artists who will bring in those consumers.

Hence, the objective of any marketing activity depends on the game on which we are playing. Now, one of the biggest objective is keep converting. Holding to your market share, gaining steadily. And that's a bigger game. Servicing it. There are different tools that have evolved on servicing efficiently, addressing them. So, one of the favorite airlines for me was Indigo. One reason they say ladies in general, boys and girls. No other airlines used to say. All the girls sitting in this room, they want to be called as girls, not ladies. So, it touches that human angle. We all love to be called boys, not gentlemen anymore. So, that human remains a human but that was a differentiation what my friend was talking about.

Are we moving into era of standardization where we were looking for differentiation? Because that's the rule of acquisition. How differentiated we are. When it is artificial, we know it's artificial. So, a lady on the television screen now, I call it television anchor Sana, has come in as a new role. We're talking multiple language. We will talk in multiple languages to you. You know it is artificial. When artificial is easily detectable, will consumer trust you when he is starting his first part of journey? Which is acquisition. I think we still as a marketeer have bigger role of using technology to convince that you are genuine.

Nothing beats the genuinity because legacy of 100 years or youngness of 5 years has no relevance till the time we are genuine to the end consumer. I don't know. Today I'm here to find a genuine closing loop because that's the biggest story and that will give success to most of the marketeers. See, remember one thing and that's what I feel. At the end of the day the role of a marketer or a brand is to reduce the consumer effort and friction. If you see these all, whether it is AI, whether it is data stacks, whether it is tech stacks, if you see them in silos, they won't make sense probably. You have to look at them as enablers. It's like me using this glass to drink the water. Can I not drink water without the glass direct from the bottle?

Take away the bottle, I can go directly jump into a pond and drink. It's about how easy you make me to consume that water. AI is like this bottle to my mind. It's not going to become the mouth. If you make it like a mouth and that's where the very, very fine line difference between AI and HI comes in. Human intelligence and artificial intelligence. Always remember as marketers that whatever you are doing and that's what when I was saying use case to your question on acquisition on wherever. Consumer cares to hoots about what is, who is serving him as long as they are getting served. We had the similar questions when website came, retailers said, oh, now people will read and buy. We are standing and selling for 30 years and somebody comes.

Remember that mall when self-service culture started coming in? Do you remember buying in, I remember my childhood going with a list to a shop owner giving him the list and he giving me a coke and I'm sitting and some mundo is trying to get everything. And today I go myself pay and become a servant of collecting everything and then bag it, pay for the bag and get it. That's where the evolution of consumer journey has happened. So it's not about acquisition, retention, any movement of truth for consumer. Irrespective of that as a brand, as a marketer, the responsibility is the tool you use is irrespective of how you are making it easy for consumption and how you are reducing the cognitive load on the consumer.

And if you think from that perspective, you will have business cases and wherever it is, wherever it's repetitive, you can automate. You can automate.