The concept of marketplace has been completely revolutionised by the advent of eCommerce. Not stopping there, eCommerce portals are taking a big leap towards more efficient, personalised, even automated customer journeys with the introduction of artificial intelligence (AI). It is the intelligence that is exhibited by machines or software, as the name suggests.
AI has found reference historically, right from appearing in Greek myths, such as Talos of Crete, the bronze robot of Hephaestus, and Pygmalion’s Galatea. But it wasn’t until the summer of 1956, that the field of AI research was founded at a conference on the campus of Dartmouth College. AI, of late, has been creeping in almost all the aspects of eCommerce. From Virtual Search and Image Recognition to Product Recommendation and Virtual Trial-rooms, AI is probably en route to revolutionising the way eCommerce portals function.
Artificial Intelligence is taking over eCommerce in a big way. Some of the areas that have been transformed by the use of AI are Visual Search and Image Recognition by enabling users to discover what they want by clicking a picture and putting it on the search bar, Product Recommendation by helping users find the best of what they are looking for. “It also helps e-tailers automate their supply chain to reduce shipment time and increase accuracy to ensure zero customer complaint,” says Shamik Sharma, Chief Technology Officer, Myntra.
Personalisation matters
“Another important area where eCommerce companies are looking to implement AI is in the area of personalisation, by crunching large amounts of data to understand buying, browsing patterns and providing each customer with a unique set of information for her/his online shopping requirements,” he adds.
Myntra has developed a smart bot that accumulates fashion-related information from across the online world by crunching a large amount of data centered around consumer demand in real time. Using computer vision and Artificial Intelligence (machine learning) on the scanned data, the platform figures out what customers love. “Those insights are passed on for implementation in designs by the lab, which puts the finished products on the Myntra app,” adds Sharma.
Apurva Dalal, CTO of Urban Ladder opines, “AI is one of the most misused and overused terms. It refers to the intelligence exhibited by machines or software and by that definition, eCommerce companies already use AI heavily.”
“Traditionally, eCommerce companies have worked mainly with unstructured text problem, but when you add the influx of images/videos/content with little to no metadata, traditional machine learning algorithms don’t do as well,” adds Dalal.
Many eCommerce portals are moving towards a marketplace model. The sellers do not give proper descriptions which affect the search, filters, and sorting. “So moving to AI-based systems will bring consistency to product attributes and descriptions. Another big area where AI is deployed is in improving the conversion by making the right recommendations to the user. Now, AI is also used in trend predictions,” says Suyajith Ali, CEO, Voonik.
Most eCommerce technology companies have invested in data science as a key computer science competence and built recommender systems on top of machine learning algorithms. Amazon, Pinterest and Netflix are all using similar machine learning concepts, but each of their recommender systems are tuned heavily for their individual use cases/verticals.
All recommender systems are not alike. There is a different set of structured/unstructured data being fed to each, different techniques of correctly extracting and analysing that data and, key customisations that are specific to a vertical that need to built into the learning system.
“So, that is the key part that all eCommerce companies will have to invest in, besides, customised data collection, analytics, learning frameworks, etc, how do you pepper in the nuances of your vertical for the recommendations to work really well for your particular vertical. Most of the aspects are now available out of the box, with open source frameworks such as Apache Mahout or tensor flow...You still need data scientists leveraging these frameworks as well as customising the individual use cases depending on the eCommerce domain,” says Dalal. “It will not be an option for eCommerce firms to use AI, which will become a core competence and differentiator in eCommerce,” adds Ali.
It’s hiring time
ECommerce portals are already hiring folks with data science competence for cracking the prediction/recommendation problem. Further more, there will be increasing demand for partnerships/M&A for startups that have teams with competence in areas of computer vision and deep learning.
“Expect more and more students to also build a competence in these areas just like how mobile app development caught on few years ago. R&D budgets will have to include the cost of learning frameworks, algorithms and computational power/storage power going forward. Good news is that things are advancing very quickly in this field and so, not everything has to be engineered from the ground up. I expect smaller startups in eCommerce soon having a level playing field with the ‘Amazons’ of the world when it comes to leveraging AI/machine learning competence,” opines Dalal.
Ali adds, “Some aspects of AI have immediate practical implications such as cataloging and recommendations — this will give immediate return on investment, and technologies required here are also comparatively mature. Some are almost there, such as trend predictions. The ones that are not yet mature such as virtual reality, drone delivery etc. will take time, but still companies will be doing R&D on them to be future-safe”.
Given the fact that AI in itself is in the maturing stages, along with the concept of eCommerce, it needs to be seen how does this romance between both revolutionise the concept of shopping.