How an Indian lingerie brand uses deep tech to better omnichannel experience

There has been a recent resurgence of omnichannel drive among companies. Although surprising that the appeal of a brick-and-mortar stores still stands in ecommerce era, companies have also relied on technology to better the experience. From artificial intelligence to machine learning, companies have relied on deep tech for consistent experience and unhindered backend to appeal to the loyalty of the customer. And Zivame has strived to achieve that, believes its Chief Technology Officer, Yash Dayal.

The Indian lingerie brand is one of the few companies at the fore front of omnichannel model. And insists on bringing Zivame experience across all platforms. “We've already implemented omni channel basics like universal pricing, single inventory and are focusing on building the next level of omnichannel experiences,” says Dayal.

Data and omnichannel intersection:

AI and ML is almost synonymous with most companies now. Dayal reiterates the old adage to begin with, "Data is the new oil." Understandably the role of data has become increasingly important and AI too has been improving leaps and bounds. The amount of data which could be processed by AI has seen exponential growth. Hence, there has been a huge change in the quality of algorithms which helps get deeper insights on data in shorter time, says Dayal.

“Data and omnichannel is an interesting intersection. It allows us to understand the customer better and provide rich experience at the point which is most convenient to the customer. So, if a customer is comfortable shopping online, we are able to give much richer experience and if prefers to walk into the store, we can use the same algorithms and give a rich experience too. We are looking at a ubiquitous experience across physical and online space,” he says.

Dayal explains there are several engines that run in Zivame. One of the interesting things to note is for every article sold there are 36 sizes possible. Therefore, there is a keen focus on what kind of inventory you need to keep and the merchandise we have to buy. “Deep algorithms look at signals from customers, analyze what they are shopping for, and use that to help build prediction models which we use to help with our merchandising,” Dayal says.

The signals range from the frequency of the product getting sold at; customer signals: How many customers have browsed a particular product. Combining these with the historical data can build a fairly good prediction model to estimate the kind of merchandise spread.

Another feature of Zivame is FITCODE which enables women to identify their size. “At Zivame, we have a huge information about various women body sizes and we use that along with other recommendation algorithms. With that information we are able to actually give the right fit to a user and approximately 7 lakh women have taken the FITCODE. And about 85% of them have found the right fit via technology,” says Dayal.

There are two mechanisms to train the algorithm. With FITCODE, Dayal says, there’s a note of an explicit signal versus an implicit signal. And this is a great example of omnichannel. “We can collect both explicit signals which can come from the stores and implicit signals based on the users that have taken the FITCODE how many have purchased. We are able to get the feedback and improve,” Dayal says.

Zivame also recently introduced a chatbot. “The first problem solved with the chatbot was that of data. We tried to understand what were the primary pain points from which customers reach out to us and a lot of those are addressed through chatbot. It is easier and faster resolution,” Dayal says.

Also, during the new normal Zivame came out with a smart audio processing in the warehouse systems. This system looks automatically at what is the pin code of the customer and if it is in green or red zone and prioritizes the delivery based on the former and how essential the item is.

Role of machine learning:

Customers at Zivame are divided into three categories: Exploratory, discovery, shop phase. Here machine learning is extremely valuable. It understands which phase the customer is in and based on that the team can build the experience best suited for that phase of the journey. Dayal says it also helps in users adapting to digital, which has been on a surge since Covid19. It also helps in personalizing the experience on listing pages, check out pages and even personalizing payment preferences.

Compared to traditional ecommerce companies—where the focus is on releasing new products—Dayal says the need here is to concentrate on the app and website. “It is about how will the product feature and work across stores and online,” he says. Another major challenge is scalability during sales. “The traffic goes up almost 10 times. This is where machine learning comes in useful as it can help detect spike in traffic and automatically add servers,” says Dayal.

Machine learning is especially helpful as the tool will learn the website or app activity and it will implement that knowledge to flag any anomaly causing activity to the system administrators.

“Technology is constantly evolving. It is important to make sure people are adopted, look at new paradigms and making changes and this spans customer experience, supply chain and future projects,” concludes Dayal.