Buying into AI for best effect - BestBuy CEO Corie Barry on how his firm is embracing the potential of a revolution in retail

In retail, the customer is, of course, always right. But here’s another maxim from Corie Barry, CEO of US firm BestBuy, that provides some food for thought:

The customer always is moving faster than the organization is.

That’s particularly pertinent when it comes to the current AI hype cycle and the positioning of the retail sector as one of those most on the frontline of adoption and piloting.

For BestBuy, AI comes in three levels, says Barry:

The first is truly the product level, what we sell to people. And our goal is to use your own words back at you to be at the tip of the spear in terms of those products that people access. That's everything from operating systems like Copilot+. That is also devices like Meta Glasses, that is whole categories that probably will come back in a completely different way, like smart home, where it actually will be a lot smarter. And that is also thinking about how platforms think ChatGPT or Anthropic start to show up on products, on things like your TVs or in your homes and then how your personality stretches across those things.

Next up is the rise of agentic shopping or what Barry dubs “the outside AI”. He explains:

How do you want to show up as people are starting to use agentic shopping?...We're just really working on, how is the customer going to shop differently in an agentic world?

That leads to some familiar tech bedfellows:

In partnership with ChatGPT, we have downloaded our whole catalog. We're also beta testing advertising with them. We're also working with Gemini and working on instant checkout with them. We're on with Wizard, which is just a shopping platform.

Dog food time

Finally there is the proverbial own dog food to be eaten. BestBuy has a number of AI implementations underway internally and operationally, says Barry, but theses are being approached with care and method, he emphasizes:

My strong point of view is you need to really think about processes that you want to re-engineer end-to-end, not just point solutions, although there are some of those too. We have completely re-engineered our call centers. I know that's a common one to say, but [BestBuy is doing it] all the way to, I have a tool on my desktop, where I can ask natural-language questions and it will filter all of the calls that came through the day before. I can actually discern what is the biggest call driver, what is happening in the Northeast, what is happening based on sentiment, all in a natural language way. That is incredibly rich data for us and for me as a leader and that came from re-engineering a call center experience.

For the firn’s dev team, the pace of change is accelerating, unlike anything Barry has ever seen:

We had a 14-week sprint we were going to run. So you could pick the right box size in a store, very important. It can't be too big, it can't be too small. And that 14-week sprint turned into a day. The code was completely written. It was deployable and we had it in two stores two days later. But you have to re-engineer how the rest of the organization works if, all of a sudden, the codebase can move that quickly.

We have six examples that are moving through the organization at that speed. So I think we're entering a world now where you're both going to have re=engineered processes, and you're going to have very quick code releases of new -- very customer-centric solutions that you can update in the moment.

The other phenomenon that he points to is that AI is moving so quickly now that the platforms are writing their own code:

Claude, as an example, literally is writing its own code base, but this is why it's not just a point solution. Yes, I can get the code, then how do I deploy it? So I need to be very quickly piloting, and that's what we did in two stores, where now they have just easy technology to say, ‘This is the box and this is the thing in front of me, okay, here is the box I would recommend’. That used to be a human kind of looking at a piece of paper, trying to judge which one would I put it in, maybe happen, maybe didn't.

The reason I like this example is that it will never just be about how fast can the code base written. It's how fast then can the risk teams look at it. How fast can you make sure that you did a little QA on it?  How fast can you get it into stores and they have a new SOP (Standard Operating Procedure) to work with it, then how fast can you deploy it across 1,000 stores, so everybody is working it?

Ultimately, what is the productivity you get out of that and how do you want to leverage that productivity against maybe a different customer issue? Nothing will be as simple as, ‘Just go re-write the code’, not in retail, certainly. You have to almost re-engineer the way the company works so that you're moving with speed and agility in a very different way.

Priorities

The hardest question for any leader right now is how to prioritize investment given the changing landscape and the evolving tools and capabilities. Barry reckons no-one has the answer here:

They're lying if they say they figured it out perfectly. I think what we're trying to say is where are those problems that are hard to unwind that we haven't had enough resource to put against, that we think could result in better customer and employee experiences? And then how do we develop small teams that can quickly get after those problems.

What's most important is having business leaders who really care, he adds:

Call center re-engineering is not done by a technology team. It's done by a business leader who really cares about what the future of our employees in that space will look like, what the future of the customer experience will look like. We've had better success putting our resources against tangled problems where we can really see the benefit to the things that [leaders] care about.

I mean, in the call center example, we've seen 50% less calls because they're getting to agents because they're getting routed automatically, and we've taken an immense amount of cost out of the model. That didn't happen overnight though, that took lots of iteration and movement.

That said, Barry is quick to note that AI investment shouldn’t just be driven by cost considerations:

I worry that all of the narrative around AI as it relates to large-scale organizations is cost savings. It is a huge enabler of growth and a huge enabler actually of even just good employee experiences. I had an employee who literally over the weekend, re-wrote and created an agent for all of our employee tools. Our employees tool website is s***. Like, you would type in exactly what you needed, and it would not bring it up. And literally over a weekend, he wrote the agent that now scrapes all that employee information. Now our employees can ask very normal human questions like, ‘I'm new here, what are the forms I need to fill out?’, and it's populating.

That agent will be able to re-purpose, to also start to solve customer support problems, because it's not like you just write an agent and it only solves one thing. If it knows how to ingest data, you can change the data it ingests and now it's solving customer problems. Some of our most powerful use cases are things like creating audiences. I can take outside signals in data. I can take all my internal data. Now I have LLMs (Large Language Models) that will mash all that data together and tell me exactly how to target someone. If they've been lapsed, I can pick them at just the right moment because I'm seeing the right demand signals, target them, and we're seeing an improved ability to bring that customer back to the brand.

My take

Our founder, Dick Schulze, founded Best Buy 60 years ago, and he laid out 4 core values for Best Buy. One of those values was learning from challenge and change.

Welcome to the age of AI in retail!