When the pandemic hit in early 2020 and consumers shifted their shopping to digital channels, big-box retailer Target felt the impact. “[We] grew our digital business by about three years in three weeks,” says CIO Mike McNamara.
To keep up with demand, “we took our Cyber Monday playbook, and we just turned it on and left it on all year round,” says McNamara.
Remarkably, the scaling was fairly seamless thanks to technology investments McNamara and his team made years earlier to position Target for success.
McNamara sat down with IDG’s John Gallant at the recent Future of Cloud Summit to discuss Target’s cloud strategy and how that’s enabling innovation and business agility.
What follows are edited excerpts of that conversation. Watch the session video embedded below for more insights.
On technology as differentiator:
Mike McNamara: It didn’t take a genius, even five or six years ago, to figure out that in the future of retail, that the winners were going to be the ones with the best technology and the best supply chains. Like in retail, clearly, location and product and price and service, these things are perennially important. But really what was going to differentiate as the world moved to digital commerce was the best supply chains and the best technology.
So, it was fairly apparent then that tech was going to be a huge part of Target’s future, which it has turned out to be, and that that was going to be way too important to outsource to anybody else. So really the whole strategy then has been about, well, if I’m going to build our technology ourselves, then I have to build a great engineering workforce and have great ways of doing work. And ultimately, we have to create a great architecture and a platform for the future. So, the strategy was all about build it ourselves; then, consequentially, that was build up the team and build the architecture out. And it has been ultimately a very successful transformation in the technology team.
On Target’s cloud architecture:
The nature of our business is very “peakish.” There’s clearly a big seasonal trend. Historically, Cyber Monday has been a blowout day, for instance: On Cyber Monday, we’re about 20 times bigger than your normal run of the mill Tuesday. So, if you build out your technology and your infrastructure to serve a Cyber Monday, the reality is that you’d end up with 90% of your assets just sitting on their ass doing nothing all year round. Cloud has been always extremely attractive to retailers for that reason, that we need the burst capacity, we need elasticity, unlike probably any other industry.
It’s also going to be important in terms of just great guest experience. And clearly, there are points of presence from the big cloud providers all over the nation and that lowers latency and improves web speed, performance, and mobile and app performance, etc.
So, cloud was a pretty obvious thing for us to want to get into. We began that journey in 2016 and completed at about a year later. And today, Google takes most of our guest-facing traffic, so our website and app, although still a lot of processing is done in our own data centers. So, what we’ve done is we’ve actually created our own deployment platform, we call it TAP (Target Application Platform). And that allows us to push workloads into GCP or into our own data centers, or indeed, down into our own stores, of which we’ve got about 1900 across countries, and all our little mini data centers, in essence.
On embracing hybrid cloud:
I would like to be in a situation where I can deploy my workloads to where they’re best matched. And whether that is our own data centers, it’s Azure, it’s GCP, it’s all of the edge… I think that is truly hybrid, when you can decide the placement of workloads depending on the need, and on the time of day and the seasonality, etc.
The reality is, is that we can operate stable workloads much cheaper in our own data centers than we can in the cloud. If you just have a straightforward, stable workload that’s going 24/7, it’s going to be financially better to actually have your own data centers, especially since we have data centers that were paid for a long, long time ago. So that makes a huge amount of sense for the stable workloads that we have. But then for the elastic workloads, placing them where they are best needed, I think that is the true nature of hybrid cloud.
On what’s next:
Spinning up new instances in our own data centers is not as simple as whipping out your credit card and going onto AWS. But that’s the journey that we’re on. We certainly have the same toolsets that manage both our internal data centers and all the clusters there, as we have in public cloud. So, we have one set of tooling that allows us to do both. But we’re still on a journey for the automatic provisioning of scalable storage and memory and processing power in our own DCs. But we’re getting there. And that will be a journey that we’ll build out over the next couple of years.