Re-inventing retail - Walmart’s Global CTO on cloud shifts, augmented reality and turning machine learning to cost-saving good use

Profile picture for user slauchlan By Stuart Lauchlan February 26, 2021 Audio mode
Summary:
Walmart Global CTO Suresh Kumar has a long tech shopping list to get through as the world's largest retailers ups its transformation efforts.

Walmart pick-up
(Walmart)

During his recent briefing on the post-pandemic direction for Walmart, CEO Doug McMillon highlighted the mistakes that other retailers had made in their evolution, citing Sears in particular as a case in point. Such firms didn’t manage to adapt and change to the necessary degree, so why should Walmart succeed where others failed. 

The answer, he said, was tech and its successful application and exploitation. The potential there is such that makes Walmart’s situation more optimistic as it looks beyond COVID, he argued: 

The thing that's different is technology. The Internet is different. Digital is different. The way you can stitch these things together is different…If we can design these things in a way that we become more of a default for certain aspects of [customers] life because of the way we've intuitively designed things, that's where the magic can really happen. And that is possible today because of digital and technology when it wasn't years ago.

As noted at the time, that’s going to come with a big price tag - 2021 CapEx at Walmart is set to rise by $3 billion to $14 billion, with the most significant spend going on the tech stack and its modernization, a task overseen by Global Chief Technology Officer Suresh Kumar. It’s work that has already begun and is paying back, he explained: 

I'm really pleased with the level of progress that we have made in this area on several different fronts. First, we added a lot of great talent to our tech team. We've brought in senior leaders from across the industry and this has really created a deep domain expertise as we start building out more modern applications. Second, we have been aggressively upgrading our infrastructure. We upgraded more than 50,000 servers, and that's allowing us to take advantage of the latest hardware and software. We also upgraded more than 2,000 stores to 1 gigabit per second fiber connection. What this is allowing us to do is we can run machine learning and data workloads, like computer vision and augmented reality, which demand a lot of bandwidth, right inside our stores. 

There’s also been a doubling down on moving to cloud platforms, he adds. Walmart has made a major commitment to Microsoft tech here, Amazon being somewhat out of the picture for obvious competitive retail reasons: 

We ran 100% of our US e-commerce and Sam's Club customer journeys on the cloud this past holiday. We also ended up building a data lake in the cloud and migrated more than 1.7 petabytes of data into it, and this is allowing us to run very advanced analytics in a very efficient manner. Lastly, we rolled out our cloud checkout system to nearly 23,000 point-of-sale devices. So this migration to the cloud really has been at the center of our modernization efforts this past year.

Scalability matters

The cloud shift proved to be a particularly prescient move when the COVID pandemic kicked in and online commerce volumes surged, recalled Kumar: 

Our every day volume level started to rise and it rose to levels even higher than our prior holiday peaks and well above anything that we had seen [since] we started running holiday shopping events. Migrating to the cloud allowed us to keep the site available for our customers, while operating lot more efficiently, because we could scale up and we could scale down in a very seamless manner.

While many online retailers were impacted by inventory fulfillment delays in the early days of the pandemic, Walmart’s cloud foundations meant that its supply chain was also able to scale: 

We lit up over 2,500 stores to start delivering online orders, in effect, turning our stores into fulfillment mode. We could do this because we built a system that crunches millions of pieces of information to find the fastest and lowest cost mode to deliver a particular order to a very specific customer. This is a huge win for our customers and for our business because not only did we deliver to customers a whole lot faster, but also at a lower cost.

The onset of COVID also drove other changes, shaped by pandemic priorities, he added: 

For example, we had to scale our VPN capabilities by 600%, our video-conferencing capabilities by over 100%. What this allowed us to do is to enable our corporate associates to work remotely without skipping a beat. Now, similarly for our customers, we delivered on features that allowed them to shop safely and conveniently, everything from contactless shopping options, COVID testing, site support, to delivery prescriptions. We launched over 100 features, big and small, to enable shopping during COVID. What I'm really proud of is how our team continued to make progress on our strategic initiatives, our tech modernization efforts, while at the same time innovating actually with speed to deliver on COVID-related business features that were needed to serve our customers and our frontline associates.

For those Walmart associates working on the frontline in stores, there has also been a tech upgrade, with innovations such as Ask Sam, a voice assistant app that can pull up store maps, look up prices, locate products and receive real-time emergency alerts. The retailer is currently testing another app, Wisbech, which uses Augmented Reality (AR) tech, said Kumar: 

What this does is it directs our associates in the backroom to very quickly identify what needs to be taken upfront. Using this app, associates are able to take only one-third of the time to complete a task which they previously had to do by scanning each and every case…We are enabling the business to move with speed, become more productive, and we are innovating on customer and associate experience.

Alongside AR, Walmart is also expanding its use of machine learning (ML), said Kumar: 

It's helping us both improve efficiency in the business, and it is transforming our customer and our associate experience. So, one good example is how we manage our assortment. We built a machine learning model to optimize the timing and the pricing of markdowns. This one effort alone saved us $30 million in markdown costs. We are going a whole lot further. We are building algorithms to better forecast demand and to optimize both the location and quantity of inventory, so that we reduce the need even to have markdowns in the first place. 

Another key area where we applied ML is in facilities maintenance. We now have an automated system to review thousands of proposals and invoices every day - and it does it for accuracy across multiple dimensions, like historical labor hours, costs of parts, travel time - and then the system recommends the ones to either approve or to reject. Our associates in real estate use the recommendations to increase proposal and invoice accuracy and this has resulted in savings of at least $40 million last year. 

One more example is on the pharmacy side where we have been building out our data lake so that we can run ML models on top of it. We took out millions of dollars of costs by improving our supplier agreements and by improving our merchandising choices. We analyzed agreements to see where we could buy a whole lot more effectively, but at the same time, we recommended what drugs could be added to our $4 generics program. This helps strengthen our customer offering, while at the same time improving our cost position…The power and the potential of ML is applicable in every single thing that we do and that is something that I'm really excited about.

Shopping list

Looking ahead, assuming the COVID crisis lessens as the Vaccine Economy kicks in, there are some specific objectives on Kumar’s to-do list: 

Last year was about building foundational capabilities, about accelerating the modernization work, to enable the business to go faster, to become more innovative, and become more productive. This year onwards, it's going to be about unlocking the future of shopping for our customers, meeting them in the shopping journey in a highly personalized omni-fashion. In fact, we want to start serving our customers right when they start consuming content, social commerce.

On our associate side, we want to reduce the time that they spend on activities like inventory counting, making multiple trips to the back room for stocking, picking, all of this kind of stuff, so that they can focus on serving our customers. We are focused on building systems that optimize all aspects of inventory and do that in real time - everything from how we get inventory from our suppliers, to keeping our products in stock, to fulfilling customer demand in the fastest and lowest cost way possible. We want to enable our merchants to focus on the art of merchandising, have the systems take care of everything else.

He concluded by emphasising the role that the tech stack will play in ensuring Walmart’s re-inventing retail long term thinking: 

The foundation that we have built is truly helping us re-invent how we serve our customers, how we run our business…Technology will enable Walmart to move with speed, be more innovative, become more productive. We see our role in technology as powering Walmart to lead the next retail disruption - and, of course, that is well under way now.