How AI-Led Personalisation Resulted in a 70% Increase in Click-Through Rates for the Automobile Platform

I would like to start with your journey: you started as a software engineer at CarWale, and today as the CEO, that is an amazing journey of 16 years. I would like you to walk us through your journey into the organisation, what you have done so far, and what your current role involves.

I am an engineer from a private college in Bhopal. CarWale back then used to be a software development company. I got an opportunity for a project here during my seventh semester, the final year of my engineering. The entire exposure since that day, including the kind of people we worked with and the challenges we have faced, has all been a great learning experience.

It was a small organisation, but the learning was tremendous. Somehow, I got lucky that the entire learning perspective, putting knowledge before everything else, actually worked for me. I keep telling my team: what worked in my favour was sticking around with the right kind of folks and putting learning before growth. Ultimately, if you are learning, you are enjoying, and growth will come eventually.

A couple of years into the job, we founded CarWale and we realised that someone needed to do content. At CarWale, we give a lot of gyaan, information, on which car to buy and how to buy, so a high amount of content focus was needed. I had some inclination towards automobiles, so I started working on content.

Then in 2008, we had worked with an agency for a search engine marketing account and they faltered big time. I was offered a small trial role: let us try for a couple of months and see what works. So we tried it, and I got that opportunity, so marketing got added to my profile, and it became more like a product and marketing kind of role.

Back then, product management was not even a phenomenon in this country. The product manager was not a role per se, but you are doing what is supposed to be done, that is, putting up a page online quickly and understanding what consumers want. Understanding how it can be grown, managing SEO and SEM, cost optimisation, and all of that.

Then eventually, there were structural changes in the organisation when we merged with CarTrade. The then CEO of CarWale moved on, the position was vacant, and I was given the opportunity. Today, as CEO, my role is to ensure that the organisation grows, the people are happy, and we continue to build our culture the way we always did.

What is the role of technology in driving the growth of CarWale, and how important is the deployment of marketing technology at your end?

The first part of the question is straightforward: in my view, we are built on technology as a company. Out of the 500 employees in the company, there would be about 170 to 180 folks on product development or some part of the technology side, be it machine learning, content building, or data science. That says it all. This is predominantly our core strength. The majority of the management team has a fair amount of exposure to technology, so it comes naturally to us. To us everything is technology first.

The world is struggling with COVID-19 right now, and everybody is cutting costs, figuring out how to survive. But we are not an organisation that is going through any kind of churn. We have not issued a single pink slip. This all comes from a simple fact that we have been an organisation that always put technology first, and that has helped us in a way.

We have always tried to be frugal. Whenever there is an opportunity to see whether people or technology can solve a problem, we put technology first. Thus we have stayed very lean, and this lean organisational approach combined with a technology-first approach has helped us sail through this, or any other problems. We have been profitable for more than two years. We are one of the few companies in the country in the online automobile sector making real profits, month on month.

The second thing which you asked about is how marketing technology helps. I will talk about our traffic here. Our business is publishing, as well as generating leads. So, it is very important that when we create traffic and generate leads, the more organic traffic we get, the lower our costs.

CarWale traffic is getting about 30 to 33 million sessions every month, and out of that, 70 to 75% on any given day is organic, either direct or Google search or referral, but we do not pay for it. We pay for the remaining 30% through Google search marketing or Facebook ads, and a few other meaningful marketing channels.

But as I said, the cost is extremely important. Through our in-house tools, approaches, measurability, and the use of the right KPIs, we would be one of the cheapest search marketing accounts in the country of our size. This means we are generating traffic from search marketing at a very nominal cost, which again helps us keep costs down. It does not happen in a day, but for sure marketing technology has helped us over a period of time.

But I can tell you that we do not do too much marketing. If you are not a car buyer, we will probably not try to lure you into our ecosystem. Keeping that in mind, when you are looking at 30 to 33 million visitor sessions every month, and considering we sell about three lakh cars in our country, I think the majority of Indian car buyers are on our portal at some point in their car-buying journey.

So, yes, technology plays a big role in general, and even in marketing, which predominantly for us is search marketing. To optimise costs and stay frugal, we use tons of in-house tools.

You mentioned how you want to focus only on the intent base. So, how are you using personalisation? Is there any AI-driven marketing, considering you are operating on a very large scale?

While the traffic is substantial, we are not a company like Amazon or Facebook, where you need to have some kind of stickiness and people hang around with you for a long time. From the 30 to 33 million visits that we get every month, typically a user visits the site 2.2 times.

You can create a cohort, and you will know that there are a certain number of people who come to you every day, maybe three times a day, but they are not the majority. When you have 2.2 visits per user, it means that people come to you once, they come the second time, and they are kind of done. They have taken a decision; they have understood what they want to buy.

However, we have had successes around AI-led personalisation, which I can tell you about. We have a team in-house doing machine learning and using an AI kind of approach for about four years now. There is a lot of investment we have done there, and I will tell you a couple of use cases that were successful.

One of our small success stories is called similar cars. Let us say Santosh comes to CarWale, and he is looking at a Toyota Fortuner. When he is done seeing a Toyota Fortuner, he would want to have a look at some options to make a thorough decision. Typically people look at two or three cars before they leave CarWale, and we wanted to make this discovery experience smoother.

Earlier our approach was: people who saw a Fortuner also saw, and then you show a few similar cars to the user. Most of it was done very mechanically, using some kind of simple algorithms based on simple categories. Then we changed that to AI-based personalisation. Our click-through rate on that particular section went up by 70 percent.

So we decided to put all this information into a machine learning algorithm. And we said, let us not decide, not put any kind of hard algorithm in, but let the system determine what people love. When they are looking at a Fortuner, do they actually look at another SUV, an Endeavour or Hector alongside, or would they rather consider a sedan, a Skoda Superb? And let it evolve, because users could have a view today and tomorrow that would change.

For example, I remember in 2011 and 2012, India was going through a churn where almost everybody wanted to buy a diesel car. And today petrol is the flavour of the day. The same goes for when SUV sales started rising in the country. Say somebody is looking at a Honda City at 15 lakh rupees; it does not mean he is only looking at sedans. If you show him an SUV alongside, could be a Creta or an EcoSport, he could be equally open to it.

So, this is one of the examples that did very well on the consumer side.

I will give you an example from the business point of view of how AI has helped us. We have been getting, in any given month, anywhere between four to five lakh leads, that we go out and further sell to dealers or manufacturers. Now we had a challenge: with the quantum of leads going up, the manufacturers and dealers started asking for some level of prioritisation or grading. So, we put in a lead scoring engine, again using machine learning. What we wanted to understand very clearly was how these consumers who convert into leads are coming on different user journeys. And if we put all of this into this engine, what came out was phenomenal. We could successfully predict who were the people more likely to buy in the immediate future.

It helped us cut down on our workforce for our calling operations. Our calling operations team, despite us delivering such a large number of leads, has only 50 people. Secondly, even for our dealers, we are building the technology to grade leads. Not just grading of who is likely to buy or who will not buy, but also, you can tell a dealer for example that Santosh was looking for a Fortuner, and by the way, he also looked at a very similar car in a similar price range.

Now, this helps the dealer put things into perspective. He can plan his sales pitch better and customise the customer experience accordingly.

These are some small examples of how we have successfully implemented AI-led personalisation.

With the auto industry in a recession, do you feel the shift to digital will increase pace? Will there be a fundamental change in the auto ecosystem?

Unfortunately, the recession is true for most sectors out there, including automobiles. Our industry specifically went through some structural reforms in the last couple of years that caused some confusion. BS4 going out and BS6 coming in took a hefty toll on this industry. BS6 cars are more expensive to produce than the BS4 equivalents. For the consumer, there is no substantial, or let us say upfront, benefit, except that my car is now cleaner and greener. Luckily most manufacturers and dealers could come out of it and were about to witness some growth. Just then the pandemic happened.

However, if I talk about digital adoption or their willingness, I think it has gone significantly up. Of course, digital marketing in the auto space was already prevalent. To give our example, I do not remember any marketer in the auto space who does not advertise on CarWale; most of them do it all the time now.

Today, all automakers are talking about how to create virtual showrooms with an online booking engine. They are preparing themselves to provide a complete digital experience for the customers. Dealers are saying, if I am not getting any kind of walk-ins, how do I make use of whatever is available? If somebody is interested in a vehicle that I sell, how can I get to them? So, there is tons of change happening, and we are a beneficiary of that.

I am not saying that I wanted the pandemic to happen. I wish it never happened, but in general, if you see the digital industry, it can come out of this sooner than, let us say, a physical sector. We are a platform where people are already there, even though we stopped some of the marketing we usually do.

The moment the country went into lockdown, we stopped some of the marketing because that was more demand-based, and the demand dried up. So, there was no point continuing to run the campaigns, but primarily if you look at traffic, it has stayed up there. People are still visiting the site and the content consumption of the portal has gone up; people are spending more time because they have undivided attention to do what they want to do.

But this is something which I believe in: when the costs usually are high, digital typically helps curtail those costs because you have measurability. When you run a print ad or show a car in the mall, measurability is all guesswork beyond a point. Digital is a lot easier: from the source until the sale, you can track everything, mapping the journey and funnel, and know what worked for you and what did not work. So you know what my cost per sale was, what exactly I made on this car, or if I lost money on this particular source.

So I think people are realising that. Some are thinking short term, that maybe I should not lose out in the current scenario; but a lot of these would realise it works and continue in the long term.

What is your observation on electric vehicles? Is India ready for it? Where are we right now, and with the pandemic, are things changing?

I think the entire EV market will take a back seat post-pandemic. Of course, these are my views, and I wish I am wrong here. But I will tell you a couple of things. Let us divide the entire EV category into three buckets: commercial vehicles, then bikes, and finally cars. Commercial vehicles would see a surge because of the pandemic.

If you do the calculation, you would realise an electric vehicle that does not run at least 3,000 to 4,000 km per month would be loss-making. It could be a 100% premium over an internal combustion engine vehicle. The main advantage of EV is cheaper fuel per unit. The second advantage is that I am reducing my carbon footprint, though this phenomenon is only applicable to a certain niche segment of buyers.

Passenger cars, I think we are more like 8 to 10 years away. It needs battery prices to come down significantly, and we would need the infrastructure to be set up on a large scale. And then, if it is a 30% premium over a regular car, I will take it. Until then, it is not going anywhere. Pushing the industry to build electric vehicles is not a good idea because the sales are not there. You would be sabotaging the industry. That is why the production of EV passenger cars is low today.

So, I think passenger vehicles, especially cars, will take a lot more time. Now bikes could adopt a little sooner, and it could be very fast with commercial vehicles. The buses, e-rickshaws, Ola and Uber and other taxi aggregators could quickly adapt to this. Because of their high daily run, they could justify the premium.

I think the industry has seen tons of challenges in the past already, and they will not want any kind of disruption for now, once things come back to normal. They would want to sell as many cars as they can to bring things back on track.