Why Companies That Are Great at Innovation Still Fail - with Stanford Professor Charles O'Reilly

**Charles O'Reilly** (0:00)
The average life expectancy of a company in the standard force 500, I think in the next couple of years will be around 12 years. So what that says to me is that the leaders of these organizations are not adapting and helping their organizations survive a long period.
We've seen companies be very good at ideation and incubation and fail. And the reason I fail is that the third discipline is scaling. Once you have a validated idea, can you actually make sure that it gets the resources it needs to actually grow and take off?
Hi, I'm Charles O'Reilly, and I'm on the faculty at the Graduate School of Business at Stanford University. I've spent the last few decades actually trying to understand why some organizations succeed in the face of change and others don't. And I'm looking forward to having a conversation about that and really about what it takes to both succeed in your existing business, to exploit existing capabilities, get better and better, but also to leverage those into more exploratory activities. And so I'm excited to have a conversation about that, and especially the role of artificial intelligence as it affects that.

**Jeremy Utley** (1:13)
Man, I am so, may I just say, I am really excited about this conversation.
This is going to be a lot of fun. So Charles, maybe we could just start with this.
The phrase explore and exploit has come up a bunch. Probably, I don't know, Henrik, what, eight times in conversations, right? It comes up often. And we thought it would be really special to have a world-leading expert on innovation strategy come and talk about, not about AI, but about what do we know from decades of research about how to help organizations innovate. And so we're delighted to get to talk to you. Just thought maybe you could start with a brief introduction to what does it mean? What does Explorer-Exploit mean? Why is it different? And why do organizations struggle with it?

**Charles O'Reilly** (2:01)
Yeah, so the term Explorer-Exploit, from an academic perspective, really came from a felon named Jim March, who is a professor here at Stanford. And he wrote a paper in 1990 that really talked about the problems that organizations face because when they get good, they get better and better at doing what is making them successful. And he refers to that as exploit. You know, we're getting better and better at our existing customers, our existing technology, existing manufacturing. But of course, when the world changes, then it requires organizations to do very different things. And that is to, you know, to look into the future, to try things, to experiment, to maybe fail, go down blind alleys, come back. And he refers to that as explore.
What that has come to, I think, mean for people who study organizations is exploitation is really leveraging your existing assets and capabilities. It's doing what is making you successful better and better, incrementally improving.
Exploration is really about what you all are experts in, which is sort of innovation and coming up with new ideas and testing new ideas. And the difficulty that organizations face is, is that the very things that make them successful in their mature or exploit business, the metrics, the incentives, the skill sets, the culture, the very things that help them be successful in their big mature business, really work against trying the things that you need to do in a world that's changing to explore. And so, that's the tension that explore and exploit captures.

**Henrik Werdelin** (3:50)
I was studying a company in China that did a interesting thing, I thought, because a lot of the companies we talk to when it comes to AI, they talk about AI as a, for, for the mechanism of saving money, right? It's that make it more efficient, fewer people.
What this company did was that they found a way to really measure, like how much money they saved with AI. And then they allocated the same amount of money into doing exploration. So they built.

**Charles O'Reilly** (4:22)
That's interesting.

**Henrik Werdelin** (4:23)
Interstudio is one of the old kind of hackathons around that. And why do you think that so few of the companies that are currently thinking about AI are thinking about it in both an exploit and an explore kind of way? Why, why lean to the exploitation?

**Charles O'Reilly** (4:41)
You know, so you all are experts in AI and of course I am not. But as I, as I think about AI, I think about it can have two very different types of effects on organizations. You can think about the substitutes and complements. That is AI can substitute for existing processes. So if we look at, for instance, law firms or consulting firms, AI or accounting firms, AI is helping them do a lot of the work that used to be done or had to be done by people. It can be done now with AI. That is a, in some ways, a substitute. A complement is it helps you do things, your existing things, but it helps you do them better. And so I think many organizations and personally as well, I don't think we quite understand whether AI is going to be a substitute or a complement. And in some instances, of course, it could be both. But the organizational response to substitutes and complements, I think, are quite different.