If there’s one high-growth company that has cracked the formula for applying machine learning to tap into the $6.5 billion lingerie market in India, it’s Clovia. Noida-headquartered lingerie e-tailer company backed by investors like AT Capital, IvyCap Ventures, Singularity Ventures, Mountain Partners AG among others was recently in the news for raising $10 million in the Series B funding round led by AT Capital along with existing investors IvyCap Ventures and private investors.
In this weekly column, we spoke to Neha Kant, Founder, Clovia to understand the algorithmic rise behind Clovia’s phenomenal growth and how the lingerie upstart is close to unseating its biggest rival in the market. As per a news report, the firm clocked ₹51.8 crore for 2017-18, a substantial rise from ₹38.7 in 2016. The bulk of the business comes from online, as Kant puts it, emphasising how the brand has grown year-on-year. Given that it is an online-first brand, 65% to 70% of sales come from the online channel. “Offline channels are also growing fast but it is also capital intensive which means the product is not only the only variable there, other factors are also coming like location, etc,” she said. Currently, Clovia operates 12 exclusive brand outlets across major cities in India. Also, the brand is a perfect case study of how to build and scale businesses using intelligent models.
Algorithms Are Driving Lingerie Business
At the heart of its operations is a robust technology backbone which has allowed it to scale operations significantly and map their production cycle efficiently. Clovia began its operations in 2013, and now ships 6,00,000 pieces per month, which means 1 piece every 5 seconds. The technology-focused online lingerie player was one of the first in the industry to build a size calculator algorithm. This was the first algorithm built by Founder & CEO Pankaj Vermani, who architected the technical infrastructure and developed a clutch of algorithms that have now become the “backbone of business”. Such is the accuracy of the model that the fit-test drives 3x conversions and the repeat purchase from Fit Test is about 80%.
Case in point, the Predictive Algorithm which gathers insights that help in production mapping and reduces the inventory cycle to just 30 days. This implies that the brand is able to sell its inventory in 30 days. Around 1,000 data points ranging from fabric to raw material, colour are gathered are crunched to understand what should be manufactured for the next cycle. “We track all of the data and decide what to make next and the production lines are mapped in a way that we are good to go. This implies that our production line is reduced to 30 days and that’s why are working capital is very organised and optimized. At any point in time, we are operating 1 to 2 months of inventory in hand,” shared Kant, who handles the sales and marketing operations at the fast-growing e-tailer.
Up next, Clovia also built a warehouse management solution in-house since the company required some specialised features which were lacking in off-the-rack solutions. Currently, the engineering team has a headcount of 16, and the staff looks after backend services, frontend, data for the app and they also have a team that builds the models from ground up. The team majorly works on Django and Python and uses MySQL for database management.
Optimising The Web Pages
Another highlight is the Recommendation Boutique where the teams tracks users after the first click is made. The recommendation boutique delivers a personalised page to a customer as soon as they select and make a second click. “We start to track their movement across the site and based on that for other people with similar behaviour we generate a personalised boutique,” shared. “We also do the same with the repeat customers by analysing the past data. So if people tend to select from the recommendation page, they are more likely to buy the product. This is all done by an algorithm,” she added.
Building An Omnichannel Brand
According to Kant, the online market share is around 3-5% which means there is a huge opportunity to tap into the rest of the 97% offline market that exists in Tier-II and Tier-III cities. The company has built a formidable database that allows it to provide granular insights according to the pincode and even different size ratios. The company is doubling down on building a national database of sorts that would help them deliver tailored collections according to the demographic’s preference. “So far, we have shipped to about 650 cities every month on a cumulative basis. Our Requirement Simulator helps in planning and creating our next production line,” she shared.
On the sales end, Kant has armed the team with an app to which also collects customer feedback. Even the network of factory partners are provided a basic app through which the manufacturing cycle can be mapped. “We make sure there is no leakage of efficiency in money and we plug all these gaps within technology,” said Kant, in closing.