Sometime last season, Fashion retailer Myntra had stocked many pairs of pink jewel studded women boots for sale but it turned out be a horrid decision as end consumers showed no interest in buying such shoes. So, the team at Myntra decided to sell all of it to their internal employees under end of season sale but even that got zero responses.
“Someone in our sales team had incorrectly predicted that pink boots would sell in large numbers but we didn’t happen to sell even a single boot to our end customers or our own employees which led to a huge waste in the system,” says Shamik Sharma, CTPO, Myntra.
Sharma did not want such precarious incidents to reoccur so he decided to look for a solution that could help them accurately forecast customer demands for every season to reduce the influx of bad products into their inventory. “This is where predictive analytics has started playing a big role. We have been able to predict people’s buying patterns correctly and it has helped us invest better every season in inventory,” says Sharma.
Myntra relies on a lot of external data sources like fashion trends from across the world, clothes colour, different types of cloth material, how people have dressed in India over the last five years and marry them with already present internal data points on the Indian market and make corrects bets on what people will wear in the coming season.
Sharma and his team of thirty data scientists not only work on demand forecasting but on a host of other problems as well. “Data science is the area which is having the most amount of impact in the business. Such predictive analytics models help us optimize the supply chain and better utilize our warehouses and help us deliver faster to our customers,” says Sharma.
Similarly, for Mumbai based real estate portal Housing.com who saw a huge surge in house searches on their site in the last one year on a per day basis were founding it difficult to extrapolate all the data related to the houses sold in terms of the location, pricing, sale history of the house and age of group of people looking at a particular house. For Housing.com, it was not practical to spend on too much manpower to find these details individually.
Therefore, Vivek Jain, CPTO, Housing.com with a team of 55 qualified data scientists would go on to build a data analytics platform using internal APIs to make sense of the unstructured data emanating from the sale of each property and make predictive models which helped them sell the right kind of property to the concerned people.
Jain hopes that through predictive analytics model built internally, he would one day be able to predict the exact kind of house each person would want to buy in India.
The Big Data Opportunity
The predictive analytics market is huge and according to analyst firm IDC, big data and business analytics applications and services will increase from $122 billion in 2015 to more than $187 billion in 2019. Add to that the $120 billion generating Indian e commerce market and you have a great platform which can be leveraged by data analysis vendors, data scientists and all e commerce platforms for much better personalized customer approach.
Sanchit Vir Gogia, Founder & CEO of the research and advisory firm Greyhound Knowledge Group points out that earlier e commerce firms used to depend on reactive analytics like push based email campaigns but over the years the need for personalization has pushed firms to use predictive analytics not just for marketing purposes but also for inventory logistics. “Predictive analytics is going beyond the consumer behaviour patterns. Firms are using cloud and writing algorithms attuned to their company needs for crunch situations like peak sales or holiday season sales,” says Gogia.
Realizing the potential of Big Data analytics, Venkateshwarlu Sonathi, Vice President – Data Sciences, Matrimony.com recollects how a certain customer of theirs spend couple of months to find her match by going through lakhs of matches available. But such a scattered search system did not yield the correct groom and cases like these led to limited customer engagement in terms of profile views, customer communication and more time taken to get the right match.
This forced Sonathi to develop a solution internally inspite of COTS (commercial off-the-shelf) recommendation solutions available which he believes do not account for the user’s demographic parameters. “We wanted to move from a “one size fits all” to a personalised recommendation engine,” says Sonathi.
This led to the formation of Intelligent Matchmaking Algorithm (MIMA) which gives suggestions real time by using mathematical rules and machine learning systems that recommends appropriate profiles to members, thereby enhancing user experience.
Cases like this are abundant. Take for example how Rajiv Mangla, CTO, Snapdeal formed a data science team of 160 members and built a predictive analytics platform using open source technology. It predicted how a mother buying diapers regularly in 2015 would need more baby products for the toddler going forward and what they should do to ensure she buys those products from them. “Personalization, in the other words, will help us drive not just repeat usage but customers for life,” says Mangla.
Even industry leaders like IBM do not shy away from admitting to the ever growing importance of predictive analytics. Anantharaman Sreenivasan, Analytics Leader, IBM India/SA says that predictive analytics matters more now, than ever before and with Cognitive coming on board, the overall efficiency of predictive decision making will enhance greatly.