Editor’s Note: This transcript was created using AI transcription and formatting tools. While we’ve reviewed it for accuracy, some errors may remain. If anything seems unclear, do refer to the episode.
Guest Introduction
Subho Mukherjee: My name is Subho Mukherjee. I’m an Associate Director EMEA at VWO and welcome to a new episode of the VWO Podcast. Today I’m pleased to be joined by Mr. Vipul Sharma, Head of Digital Revenue at SBI Securities. Vipul works at the intersection of product, customer experience, and growth at SBI, where he focuses on building digital products that are intuitive, insights-led, and impactful.
SBI Securities, which is part of the State Bank of India group, is a leading investment and financial services platform in India, supporting customers with digital-first tools for investing and trading at scale. Vipul’s work centers around improving onboarding and engagement of users while also driving sustainable revenue growth through better retention and monetization.
So with that, let’s dive into the conversation. Vipul, thank you for joining us. How are you doing on a Wednesday afternoon?
Vipul Sharma: Doing great. It was a bit busy, but it’s a closing year, so getting some more time there. But yeah, excited to deliver exceptional things in the next year.
Subho Mukherjee: Wonderful. Do you have any rituals or habits that you do to get you into the zone before a podcast, presentation, or a big meeting like this?
Vipul Sharma: To be honest, no. But what I maybe do is try to increase my confidence, boost up my morale maybe by listening to a few songs. And I believe that if I don’t care about the result, that’s when I kind of give the best output I can.
Subho Mukherjee: Got it. Are you a coffee person or a tea person?
Vipul Sharma: Coffee person.
Subho Mukherjee: I mean, coffee can be a good way to start a ritual like this. Great.
Conversation
Personalizing the First-Time User Experience
Subho Mukherjee: So we are going to start with a few questions. I would love to have a conversation broadly about your experience at SBI, and the audience would love to know about your approach — how you are thinking, not only today but also years down the line. So the first question is, when a user lands on your app for the very first time, how do you personalize their experience? Where do experiments help shape that personalization? If you can talk a bit about that.
Vipul Sharma: Sure. Maybe I’ll take that question from the time when the user creates an account with us and how we engage with them. What happens is that the first communication that you send to a user through any channel will get you the most traction. For example, an email read rate is normally around 17-20%, but the first email will garner around 60% read rate. So it’s very critical what you want to convey to the clients.
I’ll tell you the story of how we’re trying to nudge the user when they come to our app for the first time. Initially we thought, let’s maybe pitch a path that the user has to take and nudge them in that direction. We started with that and got some output, but later on we figured out that it’s not the best approach. It’s better that we don’t over-engineer the first visit — rather, we keep it simple.
Right now we ask the user what they want. We kind of showcase all the products that we have. One of the biggest advantages is that the user comes to know that SBI Securities has this many products, and second, we kind of get the preference of the customer. As I said, the first interaction gives you the highest response, so we are getting around 30-35% response on that, and based on that we’re nudging the user on how to complete that journey. The future nudges are also in line.
And it’s not only a single product. A user might select mutual funds, maybe PMS, maybe intraday. So we know what the user is interested in, and later on we can use this data to re-nudge them for other products. That’s a very effective strategy that we’ve implemented.
Tracking Intent Signals Across the Funnel
Subho Mukherjee: Got it. And after the first visit, are there any repeat behaviors or signals that help you identify if a user really intends to invest? And do you use experimentation for those specific visitors?
Vipul Sharma: Yes. So we track the behavior of users across the entire funnel — upper funnel, middle funnel, and lower funnel. The upper funnel can be reading educational content or maybe browsing our basic pages. The lower funnel would be added files or maybe going to the order page. After that we kind of give scores to each interaction of the user. So if a user is seeing educational content, it’s plus two. If they’re going to the order page, it’s plus five. Based on the overall aggregate score, we understand what is the level of involvement of the customer in the specific product.
Once we have that clarity — the frequency, the depth of usage of the product — we feed all this data into ML models where ML tries to determine the optimum score for each interaction. It also tries to understand if we need to have any capping on a particular event. So maybe a top gainers page — going five times is the cap. After that it really doesn’t add any value, so it’ll just cap the number there. And finally, when we have the overall score, we nudge the user based on that. I would want to maybe spend more or reach out to a customer through various means, including my dealer channel, my on-ground people, where the score is high. That’s how we solve this.
Subho Mukherjee: But are you running experiments around these funnels?
Vipul Sharma: Yes, yes. Once you get the numbers, after that there are a lot of experiments which go on. As a user continuously interacts but does not convert, that’s when models like these really help us. It helps us optimize cost because we have a lot of customers. WhatsApp, for example — Meta has recently revised the rates, and it’s gone upwards again. So it helps us optimize cost. And our dealers also have limited bandwidth, so giving them optimum leads is what we try to do.
Combining Behavioral and Financial Data Responsibly
Subho Mukherjee: Got it. And when it comes to investing flows within apps, specifically in the Indian market, users need guidance that has to be a blend between what they’re interested in and their financial comfort zone. My question is around how you combine these behavioral attributes that they’re showing and the financial indicators — like their engagement, CIBIL score, CIBIL data, all that stuff — to personalize the activation more responsibly, more profitably, in a more predictable manner. Is there a way that you combine these things together?
Vipul Sharma: All of this data is available to us at one place so that we can segment users and nudge them accordingly. But you’re right — investing is not a fits-all approach. I would split the entire base into three parts. One is the trading base, which is trying to make money quickly. Second is the positional base — they don’t want to make money quickly, but they’re giving it some time, maybe 15 days, a month or so. And then there are investors who want to buy and sit tight.
For each of them, each product has a different risk appetite, and we try to measure that by looking at the transactional pattern of the customer. So I know what value the customer is trading, what they’re adding, what is their total with us — plus CIBIL data, plus the KYC data. Since we’re a broker, there’s a really heavy KYC which has to be done, and that empowers us with a lot of data as well. Everything combined, we use that data to appropriately nudge the user.
However, I would also say that while we try to ensure that a person who does not have a lot of money should not come into derivatives, there are still users who do come. I don’t want to compare broking with gambling, but there’s a lot of thrill associated with it. SEBI has mandated that every broker has to show that 90% of F&O people make losses, and we do that on every login. Did that deter the customer from maybe placing the first trade or starting the journey? The answer is no, because the numbers are the same. It’s just that an extra layer has been added. While we are taking measures to do that, the pull of the customer is finally what determines what happens. So while we’ll have the CIBIL data and all of that, I know that if a person wants to do F&O, he’ll do it. He won’t wait for us.
Subho Mukherjee: Got it. And the CIBIL data, the KYC data — are you talking about the post-login experience only, or are you trying to monitor behavior pre-login, for net new visitors as well?
Vipul Sharma: Once we have the mobile number and the PII data details of the customer, yes, definitely. But before that, it’s not possible. And yes, we definitely take this into account to ensure that once an account is created and people don’t even log in on our app or website — which is again a huge drop-off, and that’s a huge industry-wide problem — we take that data into account to ensure we nudge appropriate customers so that they come to our app and ensure we do justice to the cost which is spent to get those customers.
Subho Mukherjee: My intention of the question was the same — I wanted to know about the drop-off, which is an industry issue across brokers. Are you doing something there as well, not only in your app but on your website? Are you doing anything to reduce or address that drop-off?
Vipul Sharma: Yes. If you’re talking about account open drop-off — where the account is open but the user has not yet logged into our app — what we try to do is understand who these users are and why they’re not logging in.
Doing KYC requires — I know it’s not difficult anymore, it just requires five minutes or maybe four minutes — but still the user has invested a lot of time there. They’ve done penny drop or reverse penny drop, they’ve done e-sign. So I mean, there are points where a lot of drop-off occurs. But if the customer has done all of that, why haven’t they invested? When we spoke to those customers, we realized that their priorities have changed. There are various reasons, with the major one being the priority and importance they place on investing — it’s just not there. They say they’ll do it, they’ll do it, but it’s not clear to them that this is a very important aspect of anyone’s life.
So we also take that initiative — maybe after 15 or 20 days, if the customer has opened an account and hasn’t logged in, we tell them how the markets are doing, how different stocks are doing, and kind of draw them in through actual numbers: hey, this is something you should do. We also compare this with FD and Gold, because a lot of people have money and just park it in FD or Gold. They don’t know what the other options are. And even if they know about the share market, they don’t actively participate in it. So we regularly nudge them — hey, maybe in this time period gold has given this much return. Gold has done very well recently, but at least we compare. We do that regularly to ensure we have more logins.
Key Experiments That Changed How They See User Behavior
Subho Mukherjee: Got it. Interesting. Since we are talking more around experimentation around those specific flows, I was wondering — have you run an experiment recently, or in the past few months, whether in onboarding or activation, that kind of changed the way you look at how users behave or what actually drives conversion? Has that kind of experience happened to you?
Vipul Sharma: Yes. So initially I thought that if a user’s intent is strong, users would jump through any hoop to do whatever they want to do. But I was wrong. I’ve seen that even solving small friction points — like a cluttered UI or reducing one extra click — can kill a conversion. Every small change also has a lot of impact. Especially if you’re targeting newcomers or people who are new to broking — if you have a cluttered UI, they’ll just get lost. You have to make it as simple as possible.
We have experimented by giving users various journeys — one with a simpler journey, and the second a more feature-rich journey. At least for activations, I can tell you for sure: a simpler journey works the best. And that’s maybe why, if I talk about mutual funds, why platforms like Groww will be the best journey. It’s just because they have an uncluttered UI. They’ve taken the call to remove a few things from their UI, which is a very difficult call to take if you ask me. Really prioritizing: okay, this thing we can do away with, because we know the customer needs simplicity. Maybe not providing so much customization, but giving a clean UI is a very good thing because it reduces cognitive load, which definitely improves conversion more than anything else.
Marrying ML Scoring with Experimentation
Subho Mukherjee: So reducing the number of variables — making it simpler. Got it. Earlier in this conversation you were talking about the ML scoring mechanism, which is kind of in-house to SBI Securities. And I think we had an ABFSI event as well where you spoke about the same. I’m really interested to know how you use this score as part of experimentation — what users actually see, what kind of journeys they go through, and the nudges as well. How are you marrying the scoring with experimentation?
Vipul Sharma: Sure. My team is split across different growth owners who handle different products. We have this score for each of the products and they take those numbers and ensure that if someone has high intent but has not invested in that specific product, the end goal is to ensure that they invest or subscribe to that product.
So whenever a new NFO comes in — if you’re talking mutual funds — or maybe there’s some scheme for which SIP was suspended and has just resumed, instead of just sending those details to the clients… one more example could be sending IPO details. What we’ve seen is that people who invest in IPOs, around 50% of them just come for listing gains. So if I know that a customer was shown an IPO and I know through the scoring that they’re very much interested, I’ll ensure they get details about what the listing gain was for various IPOs that happened nearby, or give some analytics around it — maybe across 15 to 20 IPOs — so they’re able to get a recall: hey, they wanted to invest in an IPO and maybe it got missed.
We also do this: if a user has seen a specific IPO and didn’t invest in it, and again if we have the listing gains or loss, we inform them directly. That has also been a very good advantage — we have seen good uplift there as well.
Subho Mukherjee: Got it. So if I’m reading you correctly, you are also trying to be a kind of thought leader as it relates to pushing people into those funnels who might be interested in IPOs, and then informing them about other IPOs, average returns, something like that?
Vipul Sharma: Yes. But along with that I would also say that we are trying to tailor the journey based on these scores. As I said before, an uncluttered UI is good for newcomers. Similarly, if I know that a customer is interested in a product, they’ll see a lot of my app telling them things about that specific product. Someone who is not exploring will get basic things. Maybe you can check that in our app — if you show interest in IPOs, you’ll see a lot of nudging happening around IPOs, from educational content to giving details about the IPOs, to giving analytics about the IPOs, and maybe gathering user feedback about IPOs and how that experience was. So yeah.
Subho Mukherjee: Got it. So kind of value-driven nudges as you go through the journey. Got it.
Quality Over Volume: The Case for High-Intent Segments
Subho Mukherjee: In our experience we have seen a lot of FinTech products or apps where growth teams really push visitors for more signups to justify the acquisition spend. In your experience, is that the correct approach? Or would you think it would be better if teams focused more on high-intent signals — to maximize value for the business overall and deliver better outcomes for users? What is your opinion?
Vipul Sharma: I would say focus should be on something which stays for 20 years rather than 20 days. However, I also have a point of view that from an acquisition perspective, many companies spend close to 2,000 to 3,000 per account. It can vary according to industry, but if you’ve done that, it doesn’t matter if you spend five more to try to nudge and get the account activated.
However, I believe it should not be a hard sell. If you’re calling the customer and even though they’re not interested, you try to force it down their throat — they’ll take it, but I don’t think that brings value. Those customers will be of low quality. They’ll churn quickly. And it’ll be difficult to justify the ROI from calling because callers have a lot of calls to make. Maybe you can go to that extreme for users where you know — yes, this is a user worth pursuing — then yes, you can maybe take that call. But bringing a lot of volume just for volume’s sake might lead to more junk accounts, which will also lead to a lot of future costs because there are a lot of hidden operational costs not visible to everyone. From servicing costs to compliance costs to communication costs — every communication we send has a cost. So there has to be a fine balance, but yes, focus should obviously be on high-intent segments, because that’s where the true value lies. That’s where you’ll get phenomenal ROI, and more than that, you’ll get a sustainable business.
AI’s Role in Activation and Personalization
Subho Mukherjee: We also believe the same — intent signals and high-intent signals are of critical importance, especially in the journeys that you’re designing. Let’s move slightly from ML to AI. How do you see AI enhancing your activation journey beyond the ML models that you’re using right now? For example, the nudges throughout these journeys, the timing of these nudges, those value-added snippets you were talking about. Are you doing anything as it relates to implementing AI there? What are your early learnings?
Vipul Sharma: Definitely. If you’re talking about what we are doing, we definitely look at the best timing, the best channel — all of that is there. And that really helps because it’s kind of a one-on-one understanding of the customer, which gives you better ROI and a better experience.
However, if you ask me how I see the AI picture coming into place — AI is far more superior than any expert, I would say, in a particular subject, and it kind of knows everything. On marketing and product, I would say that AI will help customize the experience down to the individual level. Because currently when we are doing all the customizations, we are creating segments and we have limitations. My team would have created only 10 or maybe 15 variants — 15 segments will get 15 types of nudges. But when AI comes into the picture, it’ll help segment users maybe into a thousand cohorts and determine the perfect nudge in the language they prefer and the tonality they want. This is just impossible to do manually. So AI will definitely do that, which will give a lot more conversions.
Additionally, I believe AI’s biggest role will not be in selling — yes, it’ll definitely be in selling, but it’ll be more in contextual support. I can imagine that once AI is deeply integrated in an application, it’ll notice that a user is struggling with a rejected order or facing some issues — they’re repeatedly clicking on something — and it’ll understand why that’s happening and solve it then and there itself. The entire loop that we currently have — user is trying, he doesn’t understand, he calls customer support, there’s waiting, and even if there’s no waiting, the customer support agent has to have everything explained from scratch — that takes a lot of time. And there are a lot of hidden drop-offs in this entire process.
Subho Mukherjee: Got it. For sure. It’s going to completely redefine how we think about how visitors go through these funnels and what we can improve. Like the example you mentioned — right now it might be 10 buckets or 10 different segments, and it can be a thousand segments or more, for those hyper-personalized experiences.
Risk, Safeguards, and Responsible Nudging
Subho Mukherjee: The Indian investor ecosystem is complex, as you would know. There’s a big aspect of education, and there’s a big aspect of risk. I want to go more on the risk part slightly — some of these actions, investing, derivatives, trading, some of the riskier parts, they do carry consequences. What kind of safeguards do you put into place before personalizing or nudging in those sensitive areas where you know a certain cohort might be sensitive to a certain product?
Vipul Sharma: We take a very cautious approach. Not just derivatives — it can also be stocks. If we know that a stock is very risky in nature, we ensure that we nudge the customer before placing an order: hey, this is a high-risk stock, and here’s why. Additionally, before nudging a user towards derivatives, we ensure that they have some basic knowledge or threshold.
We also ensure that if there are deep out-of-the-money options, they’re not directly visible to a new user through search, because the probability of making losses there is huge. We’ve tried to build the product in a way that if there’s a probability of a huge loss, we at least inform the customer as a cautionary step that this will most probably lead to a lot of loss — that it’ll be like eroding the entire capital — because normally people buy options, they don’t sell options.
Subho Mukherjee: And there must be safety regulations as well driving some of these actions.
Vipul Sharma: No doubt about it. SEBI is also very careful around that. What they did is they changed the expiry regulations because a lot of business for brokers happens on expiry days for derivatives — the volume spikes up. The only reason is that expiry day is a hero-or-zero day. Either you’ll make multifold money — your money will become two x, three x — or it can become zero. So most of the business was there. But SEBI has changed regulations regarding the expiry days that they can have, which significantly impacted our entire business. And they’re coming up with more regulations as well, where we have to do suitability assessments about the customer. And as I mentioned, we had to showcase the mandatory derivatives disclosure which shows that 90% of people make losses, that 28% of the money is basically paid in brokerage by the people making losses.
What Experimentation Can Do for the Indian Investor Community
Subho Mukherjee: Got it. That’s great that you’re doing all of this. Final couple of questions — I want to go slightly broader. Let’s talk about SBI Securities and your team. My understanding is that your experimentation framework is currently in-house. In your experience of the Indian investor community, taking a very broad approach, what kind of experiments or experimentation do you think will really benefit investors and help them actually grow their investing journey? Is it personalization, hyper-personalization, knowledge sharing? What would you say can actually help them grow?
Vipul Sharma: If you ask me, the biggest thing would be knowledge sharing. Investing and broking is not that complex. However, I’ve seen that maybe the friends I have from various backgrounds also do not know it inside out. And if they don’t know it inside out, expecting that a Tier 2 or Tier 3 person would know it at that depth — it’ll be very difficult for them to interpret it.
We have tried a lot to educate users, and not just us — every broker has tried to do that. I would say the best of all would be Zerodha, which they’ve done through the Varsity platform. However, if you see the total people who have seen the Varsity content or read through the Varsity pages, or saw the videos they recently released — when you just open that, you’ll realize that the first video has good views, but as you go down, the views decrease drastically. I would say by the fourth or fifth video, the views will be one-tenth of the first video. That clearly shows that users don’t have a lot of time to go through everything. It’ll mostly be an approach where they would want to learn only the things that are important and critical to them, not everything in depth.
With that, definitely ensuring that users don’t park their money in wrong instruments or wrong stocks will be a critical aspect. If we tighten that up a bit, it’ll definitely help.
What I’ve observed is there are a handful of people who invest by doing their own research. If they’re following some specific registered adviser, that is okay. But that is not normally the case. Normally they follow tips given by their friends, by random people, and they get tips like: hey, this stock will be two x in one year, it’ll be five x in one year. And saying no to such narratives really becomes difficult. So maybe urge users to ensure that they know what they’re investing in. And if that sounds difficult, it’s better to park money in pool investments like mutual funds, which is a much safer approach, or maybe in ETFs — top index ETFs — which is a much better approach than going at it directly.
Subho Mukherjee: Yeah, going it directly. The power of experimentation in educating and informing the Indian investor community can really be a big blue ocean territory.
Vipul Sharma: In a way which is consumable by them. Easily consumable. I’d say if there’s someone who can generate bite-size content very effectively, I’m pretty sure that will be a good start.
Rapid Fire Round
Subho Mukherjee: Wonderful, Vipul. Thank you so much for all of these questions — they’re very, very insightful responses. We do have a second part, which is more of a rapid fire round. But before that, any special plans for the holidays coming up?
Vipul Sharma: I’ll kind of work on my personal project, so I really have to dedicate time for that. And yes, I’ll also spend time with my wife. That’s also an important thing to do.
Subho Mukherjee: Absolutely. So it’s like a digital detox you’ll be doing?
Vipul Sharma: No, to be honest — when I’m talking about my personal projects, I’ll be building various things, so it’ll not be a digital detox for sure.
Subho Mukherjee: I’d love to know about the personal project that you’re working on. Can you share anything? If not, we’ll park it for some other session.
Vipul Sharma: Maybe we can park it for later — I would maybe want to launch it first. But I can give you a taste of what I’m trying to do. I want to explore AI to its entire potential. I’ve explored maybe the strategizing part — that’s basic — then video, generating music and songs, all of that I’ve done. But next is — I was a coder before doing my MBA and joining the BFSI space. So I am now trying to build apps, build websites, all of that. And it’s not something I want to do to earn money. I’d say it’s more to understand how to do things. So I’m building my own front end, middleware, backend — end-to-end. And using AI, it’s much easier. Things that I used to do before which maybe took me three or four months single-handedly, I’m able to do in 15 to 30 minutes. That’s a very big thing. I just want to see the capability of AI — how much can I push it, how much output can I get from it?
Subho Mukherjee: Wonderful. Okay, let’s do the rapid fire questions. There are nine questions, so we’ll get started. Rapid fire is rapid fire — you can’t really think too much.
Subho Mukherjee: If you were starting a career in CRO today, what is one thing that you would do differently from what you did in the past?
Vipul Sharma: Instead of being excited by an exciting idea and pursuing it immediately, I would rather try to dig deeper and figure out if it’s worth pursuing. Do a pain analysis, try to understand what’s the pain and what’s the gain it’ll give the user. That’ll be a much better thing to do. So yeah — prioritization would be the one thing I would work on.
Subho Mukherjee: Wonderful. Is there one thing that your non-industry friends from college or otherwise don’t understand about your job — what you exactly do?
Vipul Sharma: They don’t understand my job at all. They know I’m trying to do something digital, but they don’t understand what I’m doing. They just know: hey, something digital. That’s it. I wish they knew what I’m trying to do — experimenting, coming up with new ideas, driving digital business, changing UX, all of that. But yeah, it’s very difficult to convey that to them, especially for people who are not in the digital world.
Subho Mukherjee: Got it. Is there an Indian business professional or a product leader that you look up to?
Vipul Sharma: Yes. I look up to Sam. He’s very inspiring — listening to him and seeing what he has developed, it’s insane. However, I feel that it was a bit of luck because Google could have launched their AI publicly way ahead since they already had a lot of things in place, but due to maybe ethical concerns they weren’t able to. But anyway, an opportunity captured is an opportunity captured. I really respect the knowledge he has, how he delivers things, how he handles teams — end to end. I aspire to become like that.
Subho Mukherjee: Got it. And what’s your go-to travel destination?
Vipul Sharma: I don’t travel much, but I’d say it would be Europe. Whenever I have time, whenever I have a long leave, I tend to go to Europe. Any country is fine because Europe itself is beautiful.
Subho Mukherjee: Yes, absolutely. It’s stunning. You are very interested in the changes AI is going to bring over the next few years. What is one thing that you think AI will absolutely take over?
Vipul Sharma: I’d say agents, because I’m also trying to build my own AI agents — that’s where the real value is. In my current workflow, I’ve kind of integrated everything together, so AI can build the code, review the code, make changes, test, and everything. But if I’m able to create a lot of AI agents that can interact with each other and deliver something, I’m pretty sure the output will be phenomenal.
Subho Mukherjee: Got it. If not CRO or your current role, is there a separate profession that you would have chosen for yourself?
Vipul Sharma: I would have ideally continued to be a coder. I had that opportunity. I was working for Infosys and at that time I was handling the entire front end on a good tech line there. My friend also said that it’s a better place to be in because tech will grow. But I was like: no, I got into a good MBA college, so now I have to leave.
Maybe if you give me that chance, I might want to go there. But maybe not now, because with AI coming into place, I feel I’m better placed than before. Earlier I just knew coding. Now I know coding, product, project management, marketing — the SEO part, the SEM part as well — and on top of it I know growth and business. I’d say people who are in business right now or at the intersection of business and tech will benefit a lot from AI. Let’s hope I also benefit from it.
Subho Mukherjee: Got it. Final question: a dream or a goal that you want to achieve in the next three years?
Vipul Sharma: Next three years — I want to become a CBO of a nice firm. So yeah, let’s see if that happens.
Closing
Subho Mukherjee: Wonderful. Thank you so much, and we’ll wrap things up here for today. We thank you so much for joining us and for the deep and thoughtful discussion we’ve had. It was really great to have you on this podcast. And to everyone listening or watching, thank you for tuning in. If you enjoyed this episode, please subscribe to the VWO podcast so you don’t miss what’s next. Thank you so much.
