Interview with Meesho co-founder: ‘Users in smaller towns now comfortable buying online… we cater to them’

Meesho might not be the first shopping app that comes to people’s minds in India, but the app has seen record growth powered its focus on Tier 2 and 3 cities. A recent report Data.AI (formerly known as AppAnnie), claims Meesho saw breakout downloads in the Indian market in 2021 and record monthly active users.
But what exactly does Meesho offer and who are its customers and sellers? Indianexpress.com spoke to Sanjeev Barnwal, co-founder and Chief Technical Officer at Meesho to get a better understanding of the e-commerce platform that is primarily focusing on unbranded retail and consumers in non-metros. Edited excerpts:
What exactly is Meesho?
Barnwal: Meesho is a marketplace. We go deeper into different pockets of the country to get all the unbranded supplies out there. On one side, there are suppliers based out of major manufacturing hubs like Surat and Jaipur who l all the products on our marketplace. On the other side of the marketplace, we have resellers who are using Meesho to do their business. Around one and a half years back, this segment was a majority. We rely on a third-party logics ecosystem and don’t hold any inventory in between.
Sanjeev Barnwal, co-founder and Chief Technical Officer at Meesho. (Image via Meesho)
How is Meesho different from other e-commerce platforms? Who are its core users? What sort of growth has it seen?
Our users are value-conscious users and we operate with the mission of democratising internet commerce for everyone. Other e-commerce platforms have been able to solve this problem for users based in every tier-one city. The next wave of consumers, who are coming online, many of whom are slowly getting comfortable with spending time online, are our customers. A lot of them have started buying online as well through Meesho. People from very small towns form our core audience. We are present in more than 5,000 towns, with mostly low and middle-income segments.

Meesho has always targeted this user segment. But with the Covid-19 pandemic, the transformation that would have taken years for Indian consumers to become comfortable with buying online has happened very rapidly. It has helped us in getting good growth and a huge user base very quickly. Over the last year, we have grown 5x in terms of the number of users and all the other metrics. And because we saw this massive user base of customers coming online, we have quickly tweaked our platform in terms of the buying experience.
Until 2020, our platform was customised for resellers. Resellers who would come and use Meesho to do their own business. Even now our reseller base keeps growing every month, but overall we are seeing a majority of users becoming our end customers, buying directly from Meesho.
What technology does Meesho apply when it comes to running an unbranded e-commerce marketplace? What are the challenges of running such a platform?
In a branded e-commerce marketplace, it is easier because a brand promises certain quality, has standardisation, etc. When you go to an unbranded marketplace like us, when you onboard a new supplier, how do you find out whether the product that he’s claiming to sell is correct? If he is claiming the size to be L, is it actually the case? Quality is a very subjective metric.
When a new supplier uploads a product we look at the product metrics, but also use some of the deep learning techniques to extract information out of the image he has uploaded. From that, you can get some sense of what this product is. Now we try to expose it to the relevant user audience; not to everyone, but to a set of audiences which the algorithm realises is more likely to buy one of these products. Based on that we get some data, we also look at interactions of the consumer pre and post-purchase, the return percentages, etc.
Then we decide whether this supplier’s product is going to get more visibility. We make sure our algorithm controls the visibility of each seller while ensuring fairness. We make sure he scales only when his performance is above a certain bar.
We also go back to the supplier giving him the right input on what is that specific point where he needs to improve, etc. And that is where we have a price recommendation for these small businesses. Figuring out the right price for a product for small businesses is very difficult. For instance, how do you price an unbranded Kurti? We help them with price recommendations using data science. It’s a bunch of algorithms at play to make sure this marketplace stays fair.
Where are users coming from? How many sellers does Meesho have? What are the biggest categories?
Right now, around 5 per cent of the Indian population transacts with us. We have 4 lakh plus sellers on board. These downloads are mostly from small towns in Jharkhand, Odisha, Maharashtra, West Bengal, North-East.
The biggest category is women’s fashion, within that Ethnic, Western, etc. We have men’s fashion and beauty as upcoming categories.
What has helped Meesho with app downloads? Any particular reason it has seen a boost?
We have the lightest e-commerce app in India. Our app is only 14.5 MB as opposed to let’s say 55MB or higher. This matters for users in smaller cities and we have seen the APK size has a direct correlation with install rates. These users are just not comfortable downloading an app that is beyond 35MB, because there is some perception in mind that this just consumes too much data. Also, the kind of phones they have is mostly low-end smartphones with older versions of Android. Then network connectivity is not very stable. These are things that we have optimised a lot to make sure this app is geared towards our kind of user base.
Whenever we make or add any new feature, APK size is a top of the mind problem for our Android engineers. Our users are still not comfortable buying online or comfortable spending time on the internet. So you have to keep the experience very easy for them.
What sort of personalisation do you offer on the app?
Ecommerce apps inherently tend to be very complex. We have constantly tried to make the user experience (UX) very simple. We make sure that when you open the app, your homepage itself is personalised. We try to capture the intent implicitly based on your past interactions, based on whatever data you have shared with us, your past browsing activity and maybe a bunch of other data that we asked you to fill in during onboarding. This personalisation has helped increase the order per user up to 50 per cent. The majority of parameters that we use for personalisation are more implicit.
We show some feed based on the user’s interaction, which includes the kind of categories that he/she is gravitating towards, the kind of price category combination, etc. We use a combination of deep learning, collaborative filtering and a bunch of other algorithms to generate these feeds.
And how does Meesho make revenue?
Meesho makes money through ads as well. While we make money through ads, we make sure that our user experience is not compromised.
What about complaints of fraud on Meesho? It was reported in 2021 in the past that some consumers got products that they had never ordered.
We had issued a statement at the time. However, we have a complex fraud detection algorithm to prevent such behaviours. We can detect whether an order is genuine or not. And if we see anomalies there, it proactively blocks them. The algorithm looks at the number of orders, but also at what frequency it is getting placed.
The algorithm looks at a lot of signals. Let’s say on a given day, how many orders a user is placing, and has that increased suddenly? Is there a major anomaly in the addresses from which the orders are getting placed into? If we are seeing wildly different orders getting placed from different parts of the country the same user, it is an easy one. We use data science to give a lot of the inputs and figure out which orders are specifically not genuine and try and figure out what action to apply.