NRF 2020 - Target's CIO on the tech secrets behind Target's transformation - build features customers love, and abstract your infrastructure

Plenty has happened since my first CES review yesterday. Stuart assessed some potent news from IKEA (NRF 2020 - retail must take back control of its digital future, says Microsoft's Nadella as IKEA cuts loose from Amazon). Kurt examined the AI-for-retail implications, a big theme that definitely needed dissection.

Meanwhile, I taped a podcast with a sharp retail consultant (forthcoming), put myself through funky innovation tests like AR glasses and facial recognition mood sensors - and hunted out use cases to validate what's working. In NRF 2020 How Office Depot turned mobile adoption into an employee asset, I wrote:

Retailers are building a case for stores as a competitive edge - albeit with potentially massive redesigns/rethinks. I called it the "revenge of the store," but quickly changed that to "the revenge of the right kind of store" - fueled by better data on what customers need.

Who better to address that than Target? The numbers back it up: Target's digital revenues have grown from $1 billion in 2012 to $5 billion+ in 2019. And, as Target CEO Brian Cornell asserted, this isn't coming at the expense of stores - "It's making stores more relevant." Diginomica colleague and Target-watcher Stuart Lauchlan noted in March 2019:

The Target omni-channel transformation story is a welcome success story in a retail environment that is still claiming victims. We've written before about how retailers have had to relearn to love the store rather than flailing pointlessly after an online-only Amazon-wannabe agenda. Target learned that lesson earlier than most.

Retail CIOs have a customer imperative

But, as Stuart reported, Target's "love the store" push was also an ambitious technology play. For that backstory, who better to ask than the CIO? At NRF, I sat down with Target CIO and EVP Mike McNamara. First up: you can't serve Target's omni-ambition with an old school "systems and security" approach to the CIO role. I asked McNamara: if you have a good year as CIO at Target, what does that mean? He responded:

You have something of a customer-driven role in this circumstance - every retail CIO should have. They missed the boat somewhere along the line if they don't. Two-thirds of Target's year-on-year sales growth is digital. That is half technology, and half process.

The distinction between online and store is breaking down. That raises the technology bar:

If you look at Target's real growth, particularly if you look at the last year, on digital, it has been on drive-up, pickup, Shipt. So, these really convenient fulfillment options for our guests, that is a technology problem. Certainly, there's an operations and process piece to it, but it is about how do you make that offering very simple for a guest to opt into on the website, or even more difficult, make it very simple to do on the mobile phone? You've got four and a half square inches of real estate to play with. How do you make that a seamless operation for our store team members?

Customers expect a different store - and McNamara's team has to deliver:

What happens now, if you order online for a drive-up and delivery, as soon as you reach the geo-fence around the store, literally a car horn goes off on a mobile device that the team member has, and they go pick up your parcel, bring it out, put it in the truck of your car, and you're gone. That is a pretty seamless blend of physical and digital.

Cue the revenge of the store.

Well look, everybody's been talking about omni-channel, which is a word I hate and I've always hated, but unfortunately nobody's come up with a nicer word. I've been in retail for 25 years. For most of my career, we've been talking about omni-channel. I think Target does it right, and I think we're one of the very few companies, sizable retail companies, that do it right, where digital is operationally just integrated seamlessly with the rest of Target.

Serving the modern customer - new platform needed

And how is that driven on the back end?

The big thing we've done in the last few years on the tech side is: we now have one technology platform that serves Target commerce, whether that's digital commerce or physical commerce. If you put an item in your basket online, or you scan it through a self-service checkout and put it in a bag, you're interacting with the exact same technology. So, there is no difference. We no longer build technology for digital. We just build technology for commerce.

Point of sale or mobile apps - it's all on the same tech (and data) platform.

It's got two different screens. It's got a screen on a cash register and it's got a screen on your mobile phone. But that is literally the only difference between the two things.

In Stuart's piece, Target's CEO waxed poetic about all the different ways a hypothetical customer - in this case, a busy Mom, could get her shopping done - delivery, pick up in-store, drive-up with Shipt - and get it done fast. But those omni-options but big pressure on the tech. McNamara responded:

Yeah, we do have to be able to deliver on all those options, but we've created an architecture that can manage those things. Those offerings are far more difficult to describe to a guest than they are to code. To actually say, on a four inch screen, "Here's all the options you have for pickup."

Target's IT approach - when to build, buy, and open source

With this type of infrastructure investment, the CIO must make architectural choices with huge implications: build or buy? When to use open source. How has McNamara sorted that?

My philosophy is real straightforward. I'll buy things that don't matter to retail. I'll buy accounting. I'll buy payroll. Anything that's retail-specific, my default is that we'll build it ourselves.

And that's for the competitive edge or building your own retail tech?

Absolutely it's for competitive advantage. I don't want to have a commodity assortment planning system that's as good as the other bloke. I want to have one that's better. I'm only going to have a better one if I build it myself. So, we generally build, we don't buy. We have bought things, little things here and there, but it's a build-first philosophy, and it's an open source first philosophy.

How entrenched is open source at Target?

Everything we've built in the last four years has been on open source, whether that's the multitude of databases that are available to us, messaging systems; all the middleware is open source. Our tooling, our CICD pipeline is all open sourced. We're huge contributors to open source as well, so we're not just takers.

They are "in" on Kubernetes:

We've contributed a huge amount. We have our own version, for instance, of Kubernetes. So, we've taken Kubernetes and we've branched it because we run 1,800 stores, each of which has its own Kubernetes cluster.

McNamara's open source strategy? Free developers up to build apps - without infrastructure distractions.

I want the infrastructure to be invisible to my application developers. As an application developer, I don't want you worried about where your code is going to get deployed. I don't want you to worry about how many cores you need, how much memory you need, how much storage you need. You shouldn't worry about any of that stuff.

The developer action lies elsewhere:

I'll let the infrastructure team abstract all of that away from you, so that you, as an application developer, are just focused on feature and function. Because feature and function is what makes money.

Those features fall into two main buckets: automation/workforce productivity, and customer-facing impact.

It's the features that make us automate a process and make us more productive, or it's a feature that actually makes a guest happier and spend more money with us. So, that's where I want my engineers working. And then from an infrastructure point of view, I want the infrastructure to be abstracted so it's simple.

AI and ML - yes. IoT - on the radar. Drones in stores? Never.

I figured McNamara would be in on AI/ML as well, but I was wrong - he is "all in." To the tune of a 1,000 member data science team:

Part of my team is almost a thousand people who are data scientists or data engineers. So, I don't know exactly how many PhD mathematicians I have, but it's close on 100, at least. We have a team that is set up to devise ML algorithms.

What areas does that team address?

Those algorithms are around pricing, promotions, demand forecasting, obviously personalization, search, space management, clearance optimization. These are classic retail problems which have a new way of being solved through ML algorithms. So, we are heavily, heavily invested in creating ML to improve our business.

And has AI yielded results?

We use ML for big things and little things. Probably the biggest thing we use it for is still automated ordering. So, ordering product for our stores.

Statistic/rule-based models have been in play for years, but this is another level:

We've always had reasonable statistical models in the past, but ML just allows us to become increasingly sophisticated. I don't know an exact number, but probably about 20% of our replenishment products are through some new ML models that we've created.

We use ML everywhere. We use ML to tag photographs that go onto our website that says "This is a bikini top", "This is a bikini bottom." It's used for little things as well as for big things. So, I can see ML being fairly ubiquitous in its application, just to improve existing processes or to solve some very, very big problems.

And what about the other hot next-gen topics, as in IoT or robotics? McNamara thinks IoT is finally catching up to its hype ("It's getting close."). He sees how sensor data in stores could improve store operations, with edge computing advances making the difference. Just don't ask McNamara about drones in stores:

There will not be a drone flying overhead in a Target store in my lifetime.

The wrap - on data security and customer trust

Caveat - McNamara could see drones in more controlled warehouse conditions, away from customers. Before we wrapped up, I had to ask McNamara about data trust, privacy, and breaches. After all, Target can only maintain this momentum if it maintains data trust. I asked him about the actions taken after the 2013 breach (which was before McNamara joined Target). He says the changes Target made after that were massive:

We took a defense contractor type stance towards our data security, so we completely changed it.

That response includes a threat intelligence group that scours the dark web for retail data exploits, and a friendly cyber team that gives Target's system the once-over every week. "This is not an area that we're going to take any risk with," says McNamara.

My only beef with McNamara: I believe he downplays the challenge of getting the back end part of this right. Yes, the design elements to engage consumers matter - but so does platform, and getting out of data silos and legacy architectures. Look at all the retailers that haven't managed that. To pull it off, McNamara credits a cross-departmental approach:

I'm very proud of my team. They've done a phenomenal job. It takes us all. The stores have to be in on it. The merchants have to be in on it. The marketing team have to be in on it. It doesn't work unless everybody works together.