Amazon has hit almost every US retailer adversely. An exception has been Walmart. At over $500 billion in revenue, it remains more than twice as big as Amazon. In 2019, its share price rose 27%, much faster than Amazon’s and among the fastest in the Dow Jones index.
The stock price surge actually began in late 2015. That’s roughly the time the company embarked on a digital transformation of itself. The move has enabled the company to deal with its phenomenal scale — over 275 million customers visit its 11,300 stores and e-commerce sites every week — more effectively, and provide consumers with choices in shopping that were impossible earlier.
Core to this transformation has been Walmart Labs in Bengaluru, one of the company’s biggest R&D and engineering arms. “In the last three years, we have really evolved and grown here,” says the head of labs in India, Hari Vasudev, who moved from Flipkart, where he shaped the strategy for supply chain and built the technology stack for its logistics arm EKart.
The focus, he says, has been on two things: one, use data far more effectively with technologies like big data and AI/ML, to offer the best prices, reduce costs and do personalisation; and two, provide a great omni-channel experience — integrate Walmart’s traditional offline strength with online capabilities to create newer experiences.
One of the first projects Bengaluru undertook was a massive machine learning one, called competitive intelligence analytics or CIA, that involved looking at pricing of hundreds of retailers to ensure Walmart fulfils its promise of ‘everyday low price’ — ensuring its prices are the lowest among all. Just collecting the data from hundreds of competitors was a huge task, and then matching that with Walmart’s prices — factoring in issues like different pack sizes — was a complex machine learning problem, given that it had to be done across millions of items. This project has been a big success.
Similar technologies are also at work in the supply chain. The problems here are around how to fulfil an online order in the most costeffective way. Not all the items in the order may be available in the nearest fulfilment centre (FC). Then do you ship the items separately, or all together in a single box from an FC that has all the products? And when multiple people around an FC are ordering the same product and there aren’t enough of them in that FC, it gets even more complicated.
Suresh Shenoy, who is responsible for supply chain and e-commerce at Walmart’s membersonly retail warehouse Sam’s Club, says his team recently developed something called Taza Insights, which is helping determine which vendors, of say fruits, are delivering quality produce and which aren’t, which stores are handling the produce well, keeping them at right temperatures, which aren’t. “Providing this visibility is really helping to reduce waste,” he says.
Vasudev says an interesting data science project that they are involved in now is to determine how much to mark down a product which is close to its expiry date.
At the consumer end, AI/ML models for personalisation is a major focus. The system over time understands you enough to make appropriate recommendations, saving you time from going through thousands of products on offer. Rohit Kaila, VP of customer technology at Walmart Labs India, says it is also working very well in product substitutions. “Let’s say, you ordered a certain kind of potatoes and those aren’t available at this point of time. We actually can substitute those by an equivalent product that we believe you will like,” Kaila says.
Today, some of the biggest tech investments are going into providing a seamless omni-channel experience, an area where Walmart can sharply differentiate itself from Amazon, which is largely online. Vasudev calls it the next “technology wave” for retail.
Some fascinating projects are already operational. A customer can order online from the nearest store, and pick it up later. The pickup can happen as you drive past the store, with a store associate getting an alert and walking out with your bag. Or you can walk into the store — in case you have other items to purchase — go to an eight-foot tall tower that holds the online orders in specific slots, scan a barcode and have the machine automatically dispense your items. Several stores now have these towers, or lockers. Standing in a store, customers can check online prices, and if the latter is lower than in the store, they can immediately order from the online channel.
Customer behaviour is only convenience, says Kaila, indicating that Walmart is trying to make things convenient for customers wherever they are. So a lot of our efforts, Vasudev says, have been in bringing the technology across these two different stacks (online & offline) together.