The Personalization Problem – Segmentation, Context, and Timing

7:40 – Journey of Carsome With Personalization

Moderator – Interesting! A very different take on how you are delivering personalization. Also, Ravi, what’s your journey on being with Carsome like, how do you think shaping things are?

Ravi Shankar – So, I’m pretty new to the car industry. I’ve really learned CAR a couple of years ago. So trying to figure out the industry was tough. So one of the initial experiences I had about customer experiences is –

So I joined and then I am going around telling people that we are an eCommerce of cars, we actually deliver the car at your doorstep. We have actually asked the team to capture that moment. Team came back and said its been three weeks and we can’t get a picture.

The unique insight here is that people still like to pick up the car at an experience centre. They don’t like it delivered at home, right? So they want the experience of taking a picture with the car and finding the right time and convenient day for them to have that experience, right?

So, who takes experience at the centre? Right. So in a way, what we are trying to do is create the experience of buying a new car is always stressful, right? The first thing you think about is not really what kinda car you wanna like about me, there is anyone that I know who’s good at the task.

So what I do is brand, on top of your mind, and think about each car, and giving that experience to someone who wants to buy their car is crucial. Brands get built by sharing those experiences regardless, right? So in terms of that, delivering those messages across the journey, because, even when you decide to buy a car and all that. So you need constant communication across multiple different channels.

So, I’m not, I don’t think yet but we’re trying to integrate a lot of online and offline systems, including the bank systems to give constant communications for myself, I mean, you’ll, my car, the beginning. So I want to understand that and create an experience for everyone, right? So yeah, still in the learning stage.

Moderator – Interestingly. I’ll always, whenever I buy a car, I want to make sure that I pick up, cause this is a small story digressing from here. One of my first cars was a major European brand. As a kid, I was always dreaming of buying that car. Get to the showroom. It’s a brand-new car. Drive the car out of the showroom and start making some noise and it has to go into the workshop for a run. That was my first experience.

But your space can get heavy transactions. You might like second-hand cars. You have a budget. You have certain requirements, you apply, and you get to talk. How do you learn more about your customers and how do you change that experience for them? Is that something that you can share?

Ravi Shankar – So, definitely right. So, we center the way we do business on transparency and choice, we always think in these three buckets. Because, you need to trust a brand, to actually buy a high-value item for you. And it is known that the used car industry is not so transparent. So we wanted to change that. And then usually there isn’t a lot of, you know, we have a huge selection like so.

Understanding that customer is crucial, right? So we, I mean, being honest, it’s a new USP, we are trying, still trying to understand our customer. We’ve done a lot of research, we’ve done a lot of surveys. Now we are doing something for voice-over customers at the centers and then delivering the learnings to everyone later.

And the other way we get a lot of, uh, information to that, make sure, uh, most teams go and go through that. So it’s predominantly offline at this point because the number of transactions is low and that sample size is a small thing to get off this is behavior that we should go into to understand the business quality.

12:54 – Building Trust With Customers

Moderator – Um, I think everything that you do needs to build that trust with the customer makes the final difference. Bibaswan, properties, right? Expensive ones.

Bibaswan B – Yeah, exactly what Ravi is saying, right? So imagine if you’re researching cars, like what do you do with properties? So that’s the main thing that I learned. Uh, my state property, I come from eCommerce and so, again, like you’re trying to move a 10 Dollar cab and decide it is nothing like, unless you are part of the very select group of people who can make those million-dollar decisions in a matter of minutes, most people take months leading up to years.

So, what I have realized is that transactions in the property market are very infrequent for obvious reasons, but the research phase when a person is actually hunting for the right one, it’s insane. The frequency you are possessed with it, right? So you come back like 10 times a day, and put different filters, where is my child’s school, where is my office. So a bunch of different filters, so that the search pattern leads up to that actual decision, it starts off very broad and we have data to show that some person when he’s just in the market starting off his search, it’s very broad. And then that intent takes really qualified and like focuses on specific two, three things.

And then they get, like at that point, they desire a few things right. And fixated on this particular, on these two, three listings that I approve. So repeating a lot of that. So our personalization approach is to basically, you know, we have this long horizon, like years, the quickest ones are probably 6, 7, 8 months for the buying transactions.

For the long months, it can be more than one and a half, or two years. So we can see it ramping up and as it ramps up, we have to tailor the approach accordingly. So that can be the communication frequency. So if you have just purchased an apartment, I don’t need to cash you with three costs. It’s just payment, right? So, when we know that you are hot in the market and you need the right information to assist you in that decision, or to just get to know that something that you are looking for has just come on the market, that is where we focus our efforts on to really know what stage of that decision making cycle that consumer is on and tailor the approach accordingly. So that’s the high level that we need to.

Moderator – You also look after the B2B cycle?

Bibaswan B – Yes, I do. So my role is across the Fintech part of it, which is also very interesting, the financing, the agents, and developers. So it’s very complicated.

15:46 – Audience Segmentation At Nas Academy

Moderator – So Property Guru has definitely become a bit more complex from the way it started. And it’s been a journey though, so it’s a good success story for the region. I wanna go back to Diane because a friend Sean earlier came on board and spoke about Web 3.0. Oh, I meet a lot of people who are still figuring out 2.0. I’m probably ageing a bit so I need to catch up on 3.0 as well.

Nas Academy has a lot of courses. And before we get to the question, how many people are actively in crypto? Oh, okay. Right. So Nas Academy has a large number of courses that include a lot of crypto and NFTa, et cetera. How do you segment the kind of audience who wants to look at the schools is a very different audience than what we used to see. Then how do you manage the segregation and result?

Diane Low – That’s a great question. I think I remember people’s preferences, a point in time where they give us content emails. That’s as simple as remembering to express it and, uh, whether or not they’ve ended something to cut.

We also have something like a class that is gonna end up where people are like, I think you have an interesting website. I don’t really know what I’m looking for. So there’s like a survey about why are you interested in doing it, Would you call yourself a beginner at the end of it, a class recommendation? So, the way that we segment our customers between content creation and we have new that in the past, the bulk of our base was people who wanted to learn skill sets, like shopping on video content and becoming a YouTuber.

What’s the difference between that for travel versus like a travel vlog versus like tech reviews? Obviously, that audience is very, very different from someone who wants to learn how to invest in NFTs, like from the content they wanna hear and the view that they wanna see. And so what we do is to segment the user needs of people who are interested in content creation, business, and entrepreneurship.

And we just have three, four buckets where every week we have like more of like content newsletters where we curate from the most interesting things that happen in the space. It’s really hard for anyone to actually say that they’re experts in Crypto NFTs, even people who are teaching classes say that they don’t trust that it is the best class.

They’re very conservative with it. The space moves so quickly and what would’ve been a good investment today might experience like our schedule tomorrow. And it’s, it’s a very fasting space, the early days. But because of that, I think giving value upfront is something that’s really important.

And so, remember how to fall and understand that there is increasingly an overlap between the space of content creation. YouTube now is going up, I think the energy to enable content creators to make YouTube videos is so much like, I guess internet culture is based on online memes and lots of memorable a reference.

That all comes as to why or like because there’s something intangible that. Consumers may. So, yeah, and setting that segment allows us to write content for why beginner content creators should be cared about understanding crypto NFTs because it ultimately benefits them. The revenue sharing you get from it is much higher than the amount you make off YouTube.

I think 50-50, but obviously, you get like, think not wrong, 90. Opportunities, 2% of transaction value, two to 3%. So the remainder goes back to the creator. So if you guys are creators in the group, you should email me.

I think the benefit of what we’re trying to do is to make it understandable and as entertaining as watching a TikTok, right? So come from content creation. We’re trying to make something super pro-driven and you know, disco channels and like Twitter accessible for a person, a full-time job, just like an hour and an hour to just watch a little bit of content and have free walks to figure out how to diversify the investments and to do so and educate them. That’s just really important. Yeah.

Moderator – Yeah. I downloaded the app and I was trying to use it but I couldn’t. But uh, it’s an interesting space for you and then you are trying to educate people from that sort of expectation. So, some great learnings that, um, I think the next thing that I wanna really touch upon is also about metrics and insights. Uh, another area where I feel there’s a lot more information that can be in terms of how people are doing it in different defended aspects and cause there’s always confusion cause everything’s measurable to let’s just go bananas and start measuring, and eventually I don’t think those reports or those insights are being then used to make any changes that have been done in silos.

So Ravi in your case, how are you bridging this gap? I asked also if you have had a number of systems that I would’ve used and then tried to come up with the denominator of the customer and then we utilize it. So if you could throw some light on how nicely you’ve achieved this, whether never or generally speaking in your experience, that would be very useful for the audience.

Ravi Shankar – Sure. Just to give some context, one thing we always overlooked is the amount of behavioral data that we have. And, for a business, on average, conversion rate, the ultimate conversion rate is about 5 cents – 2 cents sometimes, right? So, we tend to focus a lot more on the transactive users, the conversion rate, and, the 9%. But the users who behave on that, on whether or we need to focus on that, right?

So that’s why, uh, the importance of the data comes in and, and that’s how you, the research why. So I tend to focus more on the transactions, not like words sold, right? What to analyze? We analyze it later, after a month or year, but how can we improve and, and all the behaviors measure the behaviors, improve their experience, right?

So even at Carsome, we have a lot of different business models, right? So we actually buy cars, consumers, and people are always listing platforms. It’s not. Uh, they come to us with the car and then they pay immediately to take the car and then they take it, you know? Right. So, and then they also buy cars from us and we also have cars where they kind of try to, and so they’re different research pieces and then making stages with the gym.

And I have a management framework, for each of those businesses, and I usually don’t give a lot of behavior, Google access to a lot of, I tend to create dashboards that each team should look at, and most of the time it’s business metrics and behavior metrics not right, not really the transactional metrics. So, I think it kinda starts from KPIs right?

So it’s a bit common when I don’t get the visibility. It happened, right? So how, trying to focus on this measurement framework, were to improve this condition rate, which is from a stack of the stand, right? So that’s how I’m thinking about in a high-frequency business and even in a low-frequency business, right? Yeah.

Moderator – And do you also have anything to measure customer satisfaction? Anything different that you’re doing?

Ravi Shankar – Yeah, what we are finding and inspecting to back the customer experience. Um, and I do have all the notifications on my phone to really understand why you have that. Right. So we have a specific budget to tackle the bad experiences because the bad experience for purchasing items should really actually, so we went and customers that are not being satisfied.

The lead amount for that perspective. They can’t pay me to finish with the Js, right? So yeah, we can measure all that across.

26:04 – Metrics That Define B2B and B2C Businesses

Moderator – Interesting. The last one, how do you do things differently? And then you can also share your thoughts on what are your metrics on B2B and B2C.

Bibaswan B – See B2B is more straightforward because there are hard metrics at the end of it when our transaction happens, we pay us money subscription. So I will just talk about it because their approach is probably really interesting for the group here. Um, so we look at it that we are a marketplace that is leading a property search, right? So it is not good if a consumer is coming back to us for too long than what we would expect them to like, expand on the search.

So that would represent the failure of the market. So we have something called match rate which is, which is like effectively the number of listings that we feel is, again, this a lot of data and research that is about that. Consumers should go through a certain number of listings before they have at least their first inquiry.

So basically they’re finding something that they like. If that is getting tested, that means that there are fundamental elements of the marketplace that are not working the way it’s supposed to. So again, that’s kinda like at the end of it. So that’s what we really wanted to talk about and optimize.

And again, there are elements of trust, listing quality, and a bunch of different things. Photographs are valid. Is the information correct? Right. Do you have all the metadata that is needed for a consumer to be confident to put an inquiry that this problem can solve? Do you know what I’m looking for? So that’s the end of it.

But leading up to it, this is at the end of the day, you know, our metric. So again, you cannot do these two teams’ matches. So, break that down. We get to more realistic levels, like so recommendations, and quality. What kind of way are you getting to get that recommendation to generate traffic on the platform?

And even beyond that, are you getting the exposure for the people who are doing that? What I spoke about in that specific stage of they’re in debt where they’re really looking for property. So are you getting the information in front of the relevant people? Is the information itself something that is good working and helpful for the consumer at that stage?

And once they get to the platform, do they have all the information that they need in the listing that says what we’re confident to inquire about if all these three things work together, we will get the right match rate that we’re looking to produce. So, something like that is, is the process that we.

Moderator – Right. Sounds a bit complicated for people to do. Do you use some sort of algorithm to make this happen?

Bibaswan B – We do. So, If, you look at it at a building block level, for a marketer or a product manager optimized for some of the things in the front, it’s not very complicated. So you are optimizing for dollars.

If somebody is getting the listing page, do they click on them, or is there a balance at that stage? So at these levels, it’s not complicated. Where it does get complicated is when we come up with a match rate for the month and that’s where we look at a lot of visitors. There’s a bunch of anonymous traffic, a lot of things happening, and we use, again, a lot of in our stack to get that done, uh, along with a bunch of in-house algos that we have again.

Moderator – Perfect. Maybe I just like to also understand why you are achieving things and you always are. You must be facing some challenges. What are those top challenges that come to your mind that you are still trying to, uh, accomplish or, or get over?

Bibaswan B – So I think the biggest challenge that I have is that with a lot of data, there’s a lot of complacency I feel, right? Because we are able to capture so much and almost everybody does it. So we have a lot of data at our disposal. So we are cutting corners in the insight process. So I’ll explain what I’m trying to say. So typically insights consist of observations about the user along with a very decent understanding of why we are having that observation, right?

So what happens now, and I also have faced this in my team and myself, that once I look at the data and find a certain segment of users is behaving in a certain way, and we made some assumptions about why that is happening and a more traditional insight validation process would, experimentation, would involve a lot of qualitative feedback, to you willing validate that that assumption is correct, right before we go with the mass market feature launch market campaign.

But nowadays I think we look at our assumptions and this is it, right? I think this is why it’s happening. So let’s just launch that happening and think about it later. So, and hence there are these issues that keep cropping up that people are not happy with some cause they’re received. Some of the features, one example of a company that does this really well, I feel, is YouTube.

So you see in the recommendation feed, they have these small snippets that come up. To be very fair, YouTube doesn’t need to do that. Imagine the amount of public stream data you view, likes, and everything. So why are they asking us? What you do want to know is the context of that recommendation.

Correct. So they are actually going from observation to act because they have said that this is a good recommendation, but they still wanna validate whether the algorithm is not missing a particular context. I think that’s the challenge that I face. I have a lot of data at my disposal. I can create segments very easily and then what moving that segment is sometimes becomes a guess on my part.

I would like to be more methodical in validating my yes. But I’m just totally imagining things in my head.

31:50 – Biggest Challenges In Personalization

Moderator – You have put it really well. It’s all about timing, right? You’ve put the challenge, you know, really thought out and I keep hearing those, and I do speak to a lot of guys who are still in the process to make this happen. Ravi in your case?

Ravi Shankar – Yeah, sure. I mean, the biggest challenge that I have at this point is localization. So localization is not just in terms of marketing, right? Just in terms of the product itself. For example, the way people shop for cars and higher, is completely different, right? So Indonesia, prefer the car monthly, and the language itself is different, that the Jaguar is different. So localizing the product and market communications. Are one of the biggest challenges it goes across the messaging attributes? High line, right? So, localizing products is the biggest challenge. I mean we have a good amount of data that kinda helps us to, and, you know, improve the product a bit.

For example, people who, uh, and he more with the low calculator in Thailand, don’t readily divide. But it’s completely different, right? So in kind, they care more about the inspection report, inspection report of the car, and the features of the car.

So, localizing that product is always challenging and the learning, and then need to, the gasoline, it’s, it’s not like any other car brand. Maybe just have one brand of five. About cutting brands and totals of cars. So yeah, that’s the challenge we may have.

Generally in most eCommerce, the transaction is completely online, and you the transaction behind as the transaction gets initiated, online offline because there are a lot of initiatives in all. So combining that test experience centers, inspections, and inspection centers, combining all that for customers, uh, it’s impossible to do or build something in all.

Moderator – Thanks for that, Diane, top challenges for you?

Diane Low – Sure. I think the interesting part about the latest tech platforms is that there are some learnings from the traditional eCommerce marketplace playbook that you can adapt. I suppose that there’s one key difference which is that because a lot of the people are teaching classes themselves, like brands with huge following the launches, you have that when they post like the YouTube video.

They have a new cost that launches can design like a huge amount of traffic coming to your site and, not all of which, that not all fans of content creators design themselves to the content creator. They are just super fans. So there could be an element of them buying access and ability to meet.

The person they look up to and take a class with. And then some people themselves follow a bit more like YouTubes show videos who want to create something and therefore want to take the class. So, um, some of the challenges we have if I were to. Is to think through which aspects of the eCommerce people can adapt and how you also have like negative 10 filters.

So if in traditional e-commerce like your targeting, so your cards and how you plug that into your so that if check out, give them reasons why this class is interesting, what people are going to. But you could earn a lot of money if you really think you are more like browsers, so that’s number one.

Number two is also the localization, academy are the two biggest markets similar to any platform the US wear venue, and therefore like a number of users who are interested in your mind and patterns and how they wanna consume content, and how they want to be updated. The brand is very different. So we see emails as a lot higher on SMS, much bigger in the US, but more mobile messaging. WhatsApp free workshops are an important way to engage in developing markets as a way for them to get users to understand the credentials and the kind of tips for what they would get if they were to buy. So, yeah, localization and figuring out sizing on the opportunity is also something that’s you know, that’s important.

That’s a bit of a challenge because we’re launching even as we’re building a system for discovery. So, I think these are the top two challenges. And then the third one is, I think if you think about it. The challenge of personalization in the earlier stage is that the teams are gonna be smaller. So justifying engineering time to add additional data pipelines needs to come with like a dollar value of value.

So, balancing out what we do manually to see what could be potential outside of revenue is something that I try to balance with what I could do product tasks and think about flexibility in our system that works not just for our product currently, but what we see as a skilled channel. So yeah, those are three things that are both the challenges as well opportunities

Moderator – Interesting challenges from you as well. But thanks so much for sharing. I’d like to spare some time for any questions from the audience.