Cars24. Not Cars……24. Here’s how data crosses that distance

Abhishek Gupta, AVP - BI & Strategy at Cars24 shows us some of those spins and helps us steal a peek at what’s inside the trunk- and it’s not just AI-video-tours or a swipe-that-app experience but also the data pistons and analytics engine oil busy at work inside.

What was it like to drive in with technology on a road that has largely been unorganised, and for years?

Yes, that was a bit tough to begin with. The used-car ecosystem is quite unorganised. Ironically, a used-car transaction is a high-value transaction, but this Rs. four-to-five lakh shopping needs the same experience as a Rs. 400 t-shirt shopping. That’s the speed and UI part. But there is also the big factor of transparency in the case of cars. A car is an emotional deal. A car becomes a family member in some ways. It is crucial to gain that trust on our platform. A few things have helped us a lot in getting it all right.

And one of them is realising and executing the truth that ‘trust is king here’. That applies to both users and dealers. We have made efforts to inject transparency and detailed information to customers and vendors with comparative analysis, real-time information and well-provided information at any stage.

Has data or technology helped there?

A lot. We invest a lot in our data capabilities. Algorithms and analytics, with a strong data infrastructure, go a long way in executing our tenets of insights and transparency.

How differently significant or challenging is it in terms of the data ecosystem your business model operates in? Like first-party data to second- and third-party data markets? Especially as a lot of your business also covers adjacent areas like insurance, scrapping, maintenance, drivers, taxation etc.

Data is at the core of our model. Our business services are a function of data sitting at the spine. Like- identifying the true value of car or predicting the price. Everything hinges on data. It is something we have seen evolving. For instance- from plain AI initiatives to AI asking questions from customers to AI figuring out condition and pricing of used-cars and all the way to recommendations and personalisation- everything is a data journey. A lot of our projects and services ride on data accuracy. Some of our business verticals gain a lot from our investments in data expertise. We cover so much- procurement, audit, inspection, financing, liquidation etc. without manual intervention and in a swift way on the platform. The entire chain is data-driven with minimal human intervention. The car-purchase journey is also enriched with hands-free assistance, AI-chatbots and real-time information.

What makes you different from Spinny or CarDekho? Is it technology- in some way?

Our approach is that of an ecosystem- and way beyond a transaction. Buying a car is like rolling a dice. With AI, we can remove a lot of guess-work and uncertainty – along with wiping away the endless haggling. A car is an important part of someone’s life and family. It is not just about a purchase or sale. There is insurance, maintenance, safety etc. so much around it. Trust is a big factor here. For us, it is a key part. Trust is not a milestone, but a journey.

What made you move towards Snowflake? What was your previous environment?

We are one of the earlier adopters of Snowflake. It has been five to six years now. One reason was we wanted to move to an optimised solution compared to what we were using then. We needed bandwidth strengths and data accuracy. Especially with the scale and complexity that Cars24 was embracing at a rapid pace and all the ups-and-downs that came with those shifts. Scalability was a constraint in our legacy infrastructure.

Can you elaborate?

Different verticals have different architectures and there tends to be some fragmentation due to this nature. Availability of data with real-time streaming nature and accuracy – then- became challenging. We needed this firm and fast foot to market products at the speed we wanted to bring in. That could not be done with maintenance hassles, heavy data pipelines etc. Snowflake takes away all that stress and also assures of data quality and governance.

Was it easy to move workloads and manage them in new equivalents? Any virtual warehousing issues or migration troubles that you faced?

There has been no problem with Snowflake and our agility, compliance and performance metrics are commendable. That said, any change, from any legacy system to something new is bound to have initial challenges. However, for us, Snowflake was not like a sales team but a partner present in every bit of the journey. It also gave near-real-time support for many issues with immediate resolution.

The ecosystem you are in – with many vendors and third parties around- would have meant a lot of heterogeneity too. How did Snowflake align with that?

It is, indeed, a challenge for any data initiative – we have multiple data sources. But Snowflake gives us a plug-and-play comfort for a variety of third-party areas and with in-house connectors- along with coverage of structured and unstructured data.

How well has the decision worked-in all these years?

Look at it from a Cars24 view and you will realise how big a use-case data-sharing is. All the back-and-forth information from user to platform (and across all other players for other business services) – has been eliminated and live-data dances well on governance, security and compliance areas too. We have also tapped a 360-degree data view with this solution and moved towards personalisation and targeting in a precise way. Real-time data has manifested so well in revenue optimisation, reporting analysis, vehicle pricing, loan disbursals etc.

Would humans stay in the loop of this evolving model- after AI’s deep entry in this space? Even when buying cars or houses etc. lean towards human interactions for many reasons?

Yes. In our ecosystem, I see it as a journey. There are hybrid teams for queries, responses, networks etc. as per various categories. We are trying to fill gaps that are still present. We will work on the combination of automation and trust.

You have introduced many new areas recently- like loan management and nearby car information. How are they structured around data?

Yes, data helps in assessing creditworthiness and giving best rates- all with the consent of customer for the data. Customisation, collaboration and consent- they all play a big role as we try to give the best value to our customers.

AI-powered videos and exploration- how effective and novel are those ideas?

When we inspect a car, there are many data points – that may not be easy to gauge on a screen. The vehicle’s sound or interiors- can be made more experiential and accurate using AI built up on the pictures and on the details that a user shares. It can, then, whip up a 360-degree view with more depth as well as remarks on possible issues and some USP areas. The AI-driven digital experience is something we are very excited about. It’s in Beta now and we will scale it up.

Would you say that vertical LLMs or models are going to be mainstream soon? In your industry too?

Data is our compass. It is a core part of our DNA. Scenarios and contexts can be different for companies beyond the general data. We would be exploring automated credit-scoring, fraud detection tools and access to under-served markets. There is so much possible ahead. We have barely scratched the surface.

What’s next?

In a start-up landscape, we have been able to excel and pivot in so many ways. Changes are happening all the time. We have been able to adapt well to so many of these in the last few years and set a strong name in so many ways with many enhancements along the way. This also has a lot of areas with technology in the ambit.