Indian e-commerce companies are still novices when it comes to Artificial Intelligence (AI) and mining big data.
Nevertheless, they are still betting big on AI, which they think could be the magic bullet that will help them offer tailored shopping experiences.
In 2017, homegrown e-commerce firm Flipkart announced the launch of an initiative called AI for India, wherein the company would develop solutions for issues like deciphering complex addresses and catching address fraud. Flipkart-owned fashion brand Myntra runs two AI-powered brands, Moda Rapido, and Here and Now.
Delhi-based Paytm’s homepage is personalised and reordered differently for each of its 225 million users, and the platform makes 20,000 recommendations per second—each of them in under 20 milliseconds.
Despite the investments, the companies have yet to make a major breakthrough in providing a customised shopping experience.
“Everybody is doing it (using AI) but there’s a lot of variation in the quality of work,” said Ishan Gupta, the India managing director for US-based skilling platform Udacity. “Easy applications like chatbots are plenty but using data science to provide better resolutions is still nascent in India.”
What exactly is wrong?
Search for a “black sequin top” on an Indian e-commerce portal. There is a high chance that the website will show you tops that are just black or just covered in sequins, or maybe something irrelevant. And even if you do find what you need, there’s a likelihood of falling prey to fake reviews and buying the wrong product.
The reason, experts say, is because e-commerce companies’ approach to AI is elementary at best. “Most companies are not able to leverage the data points and come up with solutions that can delight consumers,” said Aditya Patadia, co-founder and CEO of Turing Analytics, which creates visual search and recommendation solutions for e-commerce sites.
In addition, the quality of data gathered by these portals is sub-par. “Your AI algorithms are only as good as the amount of data that they have access to,” Shivangi Tripathi, the India country manager at Dynamic Yield, an AI-powered omnichannel personalisation technology, said.
For e-commerce majors in India, it is also nearly impossible to find large data sets that are publicly available from private entities or even the government. The ecosystem is still much too scattered and lacking harmony, Akash Bhatia, co-founder Infinite Analytics, which offers personalisation solutions to e-commerce players like HealthKart and Tata Cliq, argues.
“Most companies work with their own data silos,” Bhatia said. Removing such silos could go a long way in helping companies collate a wide range of information about the end-customer from “what they eat, to what are their music preferences, to whether they show an affinity towards travel,” Bhatia added.
Even after putting the tech into play, teaching a machine to connect the dots is an arduous task. The processes behind figuring out if the person on Facebook is the same as the one transacting at Big Bazaar still need honing.
Many e-commerce companies in India are trying to improve their AI capabilities in that direction.
At Flipkart-owned fashion portals Myntra and Jabong, big data analytics is being used to derive insights about almost every aspect of the business. ”We use AI for everything, from demand forecasting, supply chain optimisation, personalisation, and recommendation systems, to customer service systems,” Jeyandran Venugopal, Myntra’s chief technology officer, told Quartz. AI is not only used to design its in-house brands but also to price the products correctly.
To convert people browsing their sites into transacting consumers, marketplaces serve up AI-based personalised and popularity-agnostic recommendations. Machine learning (getting computers to perform tasks), deep learning (analysing unstructured data), and natural language processing (using computation to analyse human languages) play a big part in the recommendation process.
Visual search is another such personalisation tool. “It works like a virtual shop assistant by highlighting products which ‘look similar’ to the product in which the customer is interested,” said Patadia of Turing Analytics, which boasts of clients like the online marketplace fbbOnline.
To create, train and maintain efficient machine-learning processes, India’s e-commerce firms also need more AI talent. And there’s a serious manpower crisis that’s coming in the way.
Not enough talent
Though India accounts for the highest share of STEM (science, technology, engineering and mathematics) graduates in the world, as per a United Nations report, there is still a huge shortage of AI talent in the country.
“India has an extreme shortage of experienced people in this field and most of our universities are also not equipped to solve it in (the) coming years,” Patadia said. “(Silicon Valley) has required talent available at their disposal and they can easily hire fresh graduates from local universities who have studied analytics and ML in depth. This is a big differentiating factor and very hard to solve.”
When available, the standard of candidates is subpar. Most people working in data-related fields wrongfully proclaim to be data science professionals. Over half of India’s data science community holds just bachelor’s degrees. In comparison, almost two-thirds of such professionals in the US hold a master’s degree or a PhD.
“It’s difficult to separate the wheat from the chaff,” Bhatia of Infinite Analytics said about the sector’s hiring woes. His firm has a handful of its sales staff in India but their product development happens only in Boston, right outside the Massachusetts Institute of Technology (MIT) campus, where the startup was born as an MBA class project back in 2012. They mostly hire data scientists from MIT and Harvard University still because of the paucity of talent in India.
Part of Flipkart’s AI staff also works out of Palo Alto and Paytm has outsourced its analytics and fraud detection to its team in Canada.
Moreover, trained data scientists who fail to keep evolving with the advent of new technologies risk becoming irrelevant quickly. Especially in a sector like retail, where the pace of innovation is quick and many of the jobs around AI and big data didn’t even exist five years ago, experts say.
“Greater and deeper industry-academic collaboration can be one area where the larger ecosystem can invest in, even more,” said Myntra’s Venugopal, adding that Myntra has forged academic partnerships both in India and abroad.
Besides training data and employees, e-commerce firms also need to spend big computational resources to store and analyse large amounts of data. And the most important trait to imbibe at the very start of a company’s AI journey is, perhaps, patience.
“One has to remember that this area of AI and machine learning are pure research areas and results can take months if not years to materialise,” Patadia said. “If such a steady and determined approach is taken, companies can build capabilities in these areas which have very high impact potential.”