Why Your Operating Model Matters More Than Your AI Strategy

PODCAST TRANSCRIPT

Sean Woolley

I think it’s going to be incumbent for business leaders to become more technically savvy. And by that mean, as you guys know, that doesn’t mean I need to go learn how to write code, but I need to learn how to interact with different AI tools.

Brian Betkowski

Hello and welcome back to Jabian’s Strategy That Works podcast. My name is Brian Betkowski. I’m here with my friend and colleague, Ed Haines.

We’re really excited to talk about our favorite topic, which is AI, of course, but we’re going to turn it on its head today. Most of our AI conversations to date have been really about how do we bring AI into the organization and what is it going to do? What decisions we should make? How is it going to impact us? And today, we’re going to flip that around and say, “Well, what if we started the other direction and said, ‘Well, how can we maybe design an operating model and how should we design our structures and operating models in our organizations, perhaps around AI or with AI as a thing?” As opposed to right now, most of our operating models have been defined and now AI is coming into that. So now, we’re going to flip it upside down today and talk about it in that way.

Ed Haines

Right. That’s part of the process that we’re going through, isn’t it? We start with this idea of when you think about there being five stages to AI, we start with this idea of like, “Oh, I’ve got AI, I’m going to use it. I’m going to use it in my ChatGPT or Claude and as a tool.” And then you move on and use it as a process. So you put AI into process and we found, and we talked about this before, that’s where you get a lot of the benefit initially. And you do that for operational efficiency and seeing how you can use AI to improve processes. Then you look at how do you use AI to improve growth and revenue. Then the next one is how do you then look at your operating model to actually really get the benefit? And of course, the fifth one being how do you reinvent? We obviously talk about that later, but open up new markets.

But really interested to talk about number four there which is, to your point, how do we flip it around so we’re not just forcing AI onto existing processes that are made for humans, or completely mostly human, or old technology type processes but how do you actually say, “Let’s start with the start of AI and then put our operating model on top of that.” And I’ve got a couple stats that back this up.

Last year, there was a study done and basically, they said about 16% of companies are actually thinking about operating model first rather than just using AI, which is actually there’s quite a good healthy number there, but it also says 84% haven’t. And so there’s an element of opportunity there for companies. Those who have that AI-first operating model thought process have been shown to see two and a half times revenue growth and 2.4 times productivity. But most importantly, nearly over three times amount of ability to scale AI at successful rates.

And I think that’s also where we have to think about the operating model is yes, it can enable better efficiency, but it could also enable quicker and faster scale because you’re fitting AI into something that’s built for purpose.

Another stat I found here was 90% of those who are using AI, of all the people who use AI, only a third of those have really scaled enterprise-wide. And I think really that operating model is likely one of those big drivers for how and the ability to actually scale it.

Brian Betkowski

Yeah. I would say the conversations we’re having with clients seem to align to those stats. So good.

So our guest today, I’m pleased to welcome back to the podcast Sean Woolley, who runs our Dallas-Fort Worth market, is also a principal in the firm, and also is the lead of our strategy and growth offering.

Sean, welcome to the podcast.

Sean Woolley

Well, thank you so much for having me. I’m delighted to talk about this topic. It’s one that’s near and dear to my heart. I think AI and automation are critical forces in all of what we’re doing in the business world and there are certainly some pitfalls.

Ed, what you talked about, that preamble is very spot on from what I’m seeing with our clients here in DFW. This concept of it’s perfectly fine to start small and think about where to layer in automation and how. But before you really take that across your enterprise, you really need to think about what the structure of your enterprise is, how decisions are made, how information flows, how that impacts key customers.

So it’s really important to spend time on, again, understanding what can be automated and why before you make big decisions around headcounts, around how you maybe restructure your organization. But I think both are necessary, as you said, Ed, to really achieve some of those results that some companies out of the marketplace are seeing now.

Brian Betkowski

Yeah. Sean, you know we’re both engineers and we’re framework guys, and so we heard you have a framework maybe to take us through. And so why don’t we do that together? Maybe you can introduce it to us and then we’ll go through piece by piece. I think people are going to enjoy it.

Sean Woolley

Yeah. There’s really four things that we’re advising clients on in this concept of a framework. I’ll introduce all four and we can maybe double-click into some of those that are more interesting for the podcast listeners.

So the first one is, I alluded to this earlier, is this concept of planning for automation first, headcount second.

The second thing I think leaders need to do is redefine, reevaluate management layers. With what’s coming with automation, it really changes how you think about things like span of control in an operating model.

The third thing we are advising our clients to do is to think very hard about where human judgment still is truly going to matter, and there’s some places that we’re uncovering, as we work with our clients to make sure that we have that human interaction where it’s going to make the most impact.

And then the fourth part of how we’re thinking about this is this concept of shifting away from functional or departmental organization and optimization and thinking more about workflow optimization, because the handoffs that are enabled by automation and AI are much more cross-cutting than sitting just in finance, or just in HR, or just in customer service.

So happy to talk about anything.

Brian Betkowski

Let’s start from the top. Let’s start with the automation one. Yeah.

Sean Woolley

I think the old question used to be, as you’re building out a new department, or launching a new product, or even scaling a whole new area of your business, a big old question you used to ask is, who do we hire next? And that’s the second, third, fourth, or fifth question that executives should be wrestling with when they think about automation. Now, the new question is really more about what can be systemized, what can you standardize, and what can be augmented before we add any labor?

So it’s a matter of concept of first, map the workflows that you’re contemplating being automated and then really zeroing in on repeatable tasks that can be better coordinated often as things like reporting tasks. And then once you have those things figured out, it really is about then what do you do with human capacity? And for some companies that’s certain measures of redeploying current headcount or moving that headcount out of the organization but oftentimes, it’s more standing up new organizations differently.

A couple examples we can talk about if you guys want to.

Brian Betkowski

Yeah. Well, real quick, have you… There’s an old saying, which I think is much more probably applicable in today’s AI world, even though it’s been always applicable which is, “Just because I can doesn’t mean I should.”

I think we’re hearing that a lot from client, because there’s obviously so many people building so many cool things with AI and so many vendors selling really cool AI stuff. There’s no shortage of things that you can do, but how do you know when you should do them? Is there any kind of measures, or tips, or things you’re seeing out there, Sean, that people could think about?

Sean Woolley

I think it goes back to some of what I know… Brian, you’ve been helping us with some clients in DFW. It’s the concept of starting small. So this concept of proof of concept. So you can do this mapping of a broader process, think about cross-cutting, cross-departmental improvements, but do that in a laboratory setting or on a small scale before you invest tens or hundreds of thousands of dollars or millions of dollars.

And what you want to look for there are do you really see efficiency gains. So it’s a matter of doing studies and understanding what the baseline is, implementing automation structures and capabilities and then testing. Did we get more throughput? Did we get that throughput more efficiently? And you can even do things, simple business cases at the front end, and then test that the business value is showing up in a small scale and once you’re confident in that, then you begin to expand it further out.

Brian Betkowski

I think it’s a really good point because we’ve used the proof of concept term before and whenever someone uses that, you have to ask yourself, “Well, what are we trying to prove?”

And it used to be quite often that literally you were trying to prove the technology. Could the technology actually work? Could it actually integrate? Could you do… Nowadays, very rarely, are you trying to prove that you actually can do it with the technology. Usually, what you’re proving these days is like, does it feel good? After I have it, is this really the experience that I want? Is this what my customers want? What my employees want? So it’s a really good question.

Ed Haines

Well, what I was thinking when he was talking there was what you’re really actually proving is what that process journey is in that case. I start at the beginning and I go through that process and I have an end and I’m not going to just rethink my existing process. I’m actually going to start over again and think about that. And can I prove out that actually this process maybe only needs four or five steps now, whereas before it needed 20 because they were layered on top.

And so maybe that, Sean, is what you’re proving to a great degree and then those gives and those gets that get passed through that process, that’s the bit that you then automate and then layer on top of that, the people, to your point, rather than say, “Hey, we’ve got these people who have these skills, how do we layer the process on top of that?” You do it the other way around.

Brian Betkowski

A lot of people, you hear a lot of the people that are writing and speaking about this now is like, “Eliminate and simplify way before you automate.” It’s like first thing you should do, “Can I eliminate some things? Can I simplify some things from a process perspective before I go try to automate them or before I try to put a new technology in?” And so yeah, a lot of the proof that we’re doing is more business proving than it is…

Ed Haines

Do I even need it? Proof, right?

Brian Betkowski

… technology proving? Do I even need to do…

Ed Haines

Stop, start, continue is the old terms. Yeah.

Brian Betkowski

Good.

Sean Woolley

Yeah. The thing I’d add to that guys is just we’ve got a greater ability to, again, do proof of concept type testing to prove or disprove that there’s value there. So like you said, before I eliminate or add an engineer, do I know, with the automation, that throughput is what we expect so that the value’s there as opposed to, like you said, Brian, a lot of the testing you used to do for launching a new software product is so much about feasibility. Does it even work?

And with what’s happening now and from a technology standpoint, that’s less the concern and more it’s like, “Well, it’s a big investment and I’m eventually going to fundamentally change the structure of my company. Before I go do that, I better be right about, is this going to really unlock value or not? Is it going to continue to delight my customers? Am I going to make and bring better products to market or those customers going to have better service experiences, etc., etc., etc.?” And that’s just the front office stuff. You can take this whole conversation and look at what’s going on with back-office activities like the human resources function.

Brian Betkowski

Yeah. I think, Sean, you’re bringing it up is also another thing that we’re seeing, which is that the value of the thing is often one hop or more away from the implementation. So the example is if you’re trying to help a salesperson be more efficient by taking some rote tasks away from them and automating those, which are really the underlying value hypothesis is with that time, they’ll now be able to do something else that’s way more valuable.

And so you go do a POC about the technology to take away the rote task and then you implement the technology, which takes away the rote task and then it’s like, “Okay, so now I have to be able to have a way to perhaps enable them to do the other task more, measure the way to do the other task more.” It immediately becomes more about the other task than it was about the POC for the rote task that you automated.

Ed Haines

Right. And the other example of that is if you had, say, a simple process where you created a document. In fact, we were talking about this on the way over here previously earlier, which is I have a process where I create a PowerPoint, let’s say, and then you go, “Okay, end of process,” but then what do I do with the PowerPoint? If I have another process that’s AI-enabled, then what’s the point in putting a PowerPoint back into it? You just put the JSON that you’re creating or whatever it is, the prompt that you’re creating into the next AI system.

So having an understanding of how all of the processes and the ecosystem work together is likely where you need to start, which is why we’re talking about operating model rather than process.

Brian Betkowski

Yeah, yeah. That’s a good point.

Sean Woolley

Yeah. This is a really good example and a nice segue into the second part of the framework I talked about earlier, which is this concept of redefining management layers.

So to follow your example further, Brian, around enabling a salesperson with less rote tasks, when you think about an operating model decision that you might make once those sales folks have better tools, think about the sales leader or the VP over a set of sales leaders, the amount of time that those middle managers once spent on understanding what his or her sales team is doing, that management of that team fundamentally changes now because as a middle manager of a sales team, I have to spend so much time chasing things down, understanding the productivity of my sales force. We can now build and put tools in front of those sales managers that literally their span of control can double or triple in some cases where I used to effectively manage a sales team of 10 or 15.

Now, with these tools that we’re building, that becomes double or triple that number. So there’s an immediate efficiency saving or cost savings for an organization, enterprise wide, because you just simply don’t need that many sales managers.

Same example we see with one of our clients in the medical device world now is what we’re helping them build is going to take a lot of either redeploy a set of, they call them, HR business partners. So these are leaders who work with this global organization to understand what HR needs do they have transactionally with all of their employees. What we’re going to enable them with is more self-service tools so that they’re not needing so much interaction from these, again, HR business partners, and that becomes a huge cost savings where that capital can be redeployed. You get it for other more productive activities.

Ed Haines

It’s actually what you’re also making me think of there is from a personal experience, as you go up the organization, the higher up you get, the more people you have, the less you touch technology and in some ways, because your job is to be strategic and to connect with people and so on, it actually is starting to turn around now where you actually may have all these agents and you really should have your hands more on the technology. I don’t know if you’ve seen that, Sean, but you might also get a bit left behind if you use the old ways.

Brian Betkowski

You’re bringing up just an interesting point about leadership versus management of the leadership parts diminish as you have to… Do you really lead AI agents? It’s an interesting semantical question like, “Are you leading them or are you just managing them?”

Ed Haines

You manage individual ones, you lead groups.

Brian Betkowski

No. It’s so interesting. It does this because all the traditional rules about how many people you could efficiently manage, how many humans you can efficiently manage to be able to do right by them from a career perspective and from their job perspective, that gets really shook up now and you can obviously lead or manage a bigger set of…

Ed Haines

But if one of your agent… We’re now stepping into a bit leadership group, but one of your agents starts to be more of a leadership assistant or a communications coach, you end up… Yeah, I suppose you’re using technology, but you’re always using technology to a certain degree, but it’s just more… I think some of these AI stuff feels more in-depth and… Yeah. Interesting.

Sean Woolley

The way I think about that guys in terms of leadership and management is when done right, I personally, as a leader of leaders, spend less time managing and more time inspiring those leaders and thinking about the strategy for the department or the strategy for the DFW market. I’m able to free up time to do the things that, again… So the third part of the framework here is evaluating where judgment really matters. This is a good example of I’m going to rely on judgment and how I inspire a fellow executive more so than I’m going to rely on an AI agent.

Now, I may use AI to help me prepare for those kinds of discussions but at the end of the day, what I’m really going to say, how I’m going to say it, how I’m going to get that leader to go run through a brick wall for me and his or her team, I can’t outsource that to AI, but now I’ve got 20, 30% more time to think really hard about how that message is going to land, when to do it. I have more opportunities to deliver those messages, really frees you up as a leader if you’re doing a little bit less management or you’re doing management more efficiently.

Ed Haines

Yeah. I think it’ll be a challenge, especially, and we’ve seen this with clients already, where you can have a manager or a leader and they can have five direct reports and then 20 agents that have reporting to them. And so when your organization ends up being full of agents, what does that look like? Because you got to be able to… Yeah, you’re not going to inspire them because they’re not there to be inspired, but you do need to… There’s a management aspect, but there’s also how do you collectively… I think there’s another layer there. I’m not really quite sure what that is yet.

Brian Betkowski

I wonder… What you’re bringing up is, even just the term span of control, that term, obviously, probably originated from something in military world of truly controlling what somebody’s doing, but it’s definitely probably more human thing.

You almost think that it’s actually changing to more of span of outcome or span of productivity. How much productivity is underneath you? The fact that sometimes a machine is doing it makes it different than talking about your org structure of how many people are in your org and now, it’s more of like, “What are you truly responsible for? What is the capability and productivity that you are producing?” Now, who’s helping you produce it? In some cases, more humans, less humans, whatever but I wonder if that’s where we’ll start talking about it that way more, as opposed to the traditional way of talking about how many people are in your organization.

Sean Woolley

Yeah.

Brian Betkowski

Interesting.

Sean Woolley

It’s funny, Brian. I’m speaking on an executive panel this evening in Fort Worth and the point of the panel, we’ve got some civic leaders, some university educational leaders, and some business leaders talking about preparedness of the workforce, and that is something that will come up tonight is as AI automation continue to infuse itself in the marketplace, I think it’s going to be incumbent for business leaders to become more technically savvy.

And by that mean, as you guys know, that doesn’t mean I need to go learn how to write code, but I need to learn how to interact with different AI tools. And I can’t farm that over to IT or my digital partners. I personally because I will have, like you said, very likely a set of agents that work for me, I need to understand more about how those agents work. So that skillset is going to become table stakes.

We’re already thinking about it at Jabian, again, how we hire, that this is no longer optional that, A, you’re excited and interested in this but, B, you’re also very capable at applying and using these tools in your day-to-day job.

Brian Betkowski

I was at a dinner last night, two CEOs, both amazing and one of them is a very culture focused, very engaged leader. And he actually asked a really interesting question, or a statement, which is more of like, okay, in today’s world, he relies heavily on the ability for his leadership team to cascade messages appropriately. He’s a very, very leadership focused, very culture focused person. So in his mind, he’s like, “That’s what I do. I set strategy, I make decisions, I cascade that down through my leaders.” And he asked a really interesting question. He’s like, “Well, what if I have to rely at some point on somewhere in that chain there being AI agents that need to…

Ed Haines

Instructions need to be on you.

Brian Betkowski

… cascade my message correctly in the way that I intended?” It was a very interesting human question.

Ed Haines

Well, yeah. Rethinking how you’re interacting in your… When we’re going back to operating model, but yeah, that’s a really good example.

Brian Betkowski

Yeah, that’s an interesting thought experiment.

Sean Woolley

Yeah, yeah, for sure.

Brian Betkowski

So what’s number four, Sean?

Sean Woolley

The fourth component is getting away from departmental and siloed thinking and thinking about value stream. So an example of a value stream is something like quote to cash.

I think traditionally we would go through as a management consultancy and say, “Okay, let’s look at quoting and optimize that.” Work really hard with the sales team and move on to order entry and again, think about sales support and how all that’s working and optimize product delivery, get to invoicing and cash collections. And you could probably name the departmental function for each of those.

If you really want to get the most out of AI and automation, you’ve got to go across all of those functions and think about the process as, again, a value stream from do we have the right customers? Are we getting the quotes that they need? Are we collecting all the right information in the right times and handing that off to delivery? And then when we do deliver our product and service, do we have the information we need to appropriately invoice them? Because if you optimize any of those in the silo, you’re going to leave money on the table, so to speak.

So it’s really about… When you think about automation, there’s a ton of value in identifying friction points in handoffs from one department to another. And again, doing that redesign end to end.

And then you got to really think about depending on what you’re doing, you could really shift workload and that has an impact on, I think, incentives and associated outcomes. So we got to understand the outcomes of the value stream, not the outcomes in a given function and reward or create incentives just in a function because then you suboptimize the outcome if you don’t think about, again, customers, products and the overall market, again, across that value stream.

That’s not new. We’ve done that before but I think with automation, there’s just so much greater ability to think about, again, this concept of friction points and handoffs.

Brian Betkowski

Yeah. It’s so funny. As you were talking, you can imagine saying that to a human, “All right, you got to take into consideration this and this and this and this.” And someone would say, “Well, that’s a little overwhelming. I’m not sure I can take all that into consideration.”

But now with AI, the context that it can handle at one time, which that’s growing exponentially almost by the day about how much context that it can keep and take into consideration at one time. And you can obviously layer on other technologies to give it even more context. But it starts to say that trying to solve larger problems like that in an efficient way is so doable where there were other perhaps barriers to that in the past.

Ed Haines

I’ve seen big organizations constantly reinventing themselves and how the departments are structured to squeeze more out of revenue or efficiency, and just experimenting with the right mix, that’s very common. And now if you take away the department level a little bit and really get down to the core gives and gets in the process, you can actually… That goes away a little bit and you make the department piece a bit thinner.

But then where my mind went, Sean, on this was, well, that’s like a shared service. Now if you move things to a shared service, do you take away the accountability of the actions of the agent within that department? So I suppose the question is, are you seeing any of that in the market, Sean, in terms of how people are thinking about where agents and AI lives? Does it live in a centralized place where you can have a little bit more flexibility around shared services, or does it live in a more siloed departmental groupings but where you keep the accountability?

Sean Woolley

I think what we’re seeing, it really varies. It depends on what the stakes of whatever the agent are handling. So I can give you a specific example.

You can think about, again, the medical device world, this concept of customer complaints. Today, there can be complaints around did I get my physical inventory, if it’s like medical device, to the warehouse or to the hospital or to the doctor’s office. There’s a bunch of, you can imagine, customer service queries that manage around that. There… We’re encouraging our clients to lean in with automation because the stakes of something being off are not incredibly high.

Now, flip the switch and it’s I have a surgical product and a doctor’s calling to tell me about a product failure software or hardware that led to an adverse event with a patient. There, I think you need to be very careful about how much human interaction comes into that conversation and how you would structure those two different operating models to what the, again, accountability and incentives, driving outcomes would be fundamentally different and I would and we have designed different operating models that have a greater reliance on that patient safety impacting side of the equation. It’s fundamentally different and that’s where human judgment needs to remain.

Now, again, you can do a lot to help those humans operate more quickly and efficiently, but final decisions will remain with a human being. Many of them. Now, do I expedite or slow down a shipment of tongue depressors? That’s different. I can let the agents have a little bit more degrees of freedom to make those kinds of decisions.

Brian Betkowski

This has been really good.

So I know just in having this conversation with a couple of executives, sometimes that term operating model, you get wrapped around with semantics on that, or sometimes you can feel a little, I don’t know, consultanty or whatever. Is there any way, Sean, just help folks just maybe boil that down or simplify, or when they hear the term operating model, what should they think?

Sean Woolley

First, don’t overthink it.

There’s a handful of elements when we talk with our clients about operating model. The first is structure. That’s fairly obvious. Another one is within structure spans and layers from an org structure component.

Another big one is information flow. So it’s very important to understand how does information move from different elements of your structure and your operating model.

The third one is a lot about how are decisions made and who makes them under what conditions. Sometimes, we call that governance but it really is about, again, how our decision is made. And where that gets really important is things like capital intensive companies, got to make a hundred million dollar decisions about where to build a factory, for example, which all again has implications on how the company operates. So who’s got the decision rights to make that big investment decisions that lives within your overarching operating model?

Another big component is just reporting and dashboards. So not just how does information flow, but what are the tools, templates, operating norms that push and pull information throughout and within your operating model? That’s as far as I would take it.

Brian Betkowski

Good.

Sean Woolley

And again, don’t overthink it but again, get it on a page. When we do operating model design, it’s very visual and get it on a page, get it on a big whiteboard so that it’s very clear. And again, that’s where you can start to pinpoint these friction points that create an opportunity for efficiency.

Ed Haines

And then you’d want to add the purpose and the strategy of the company or the drivers of that, which is why …

Sean Woolley

Yeah. And that’s one where I think for smaller, mid-size companies, some of that will fundamentally change. But when you think about Fortune 10, Fortune 20, they’re not rethinking their core purpose and their vision very much. They might tweak it or modify it. But we get into those aspects of operator model discussions a lot with our smaller or medium size entities.

Ed Haines

And also to that point, you can have an operating model for the whole business, obviously, but you can also have operating models at a smaller level that can work on their own to a certain degree, can’t they? So if anyone listening to this says, “Well, I can’t redo my whole operating model. I don’t have the ability to do that.” But you do have the ability to look at your own operating model within your own department.

Sean Woolley

For sure. The thing I’d add to that is if you’re going to really try to multiply your effect with automation, sure think about your operating model and do what you can that’s within your control but don’t do it with blinders on. If I’m the head of operations, I can’t go tell the head of sales to fundamentally re-engineer his or her operator model. But if I’m going to do something over here, I need to at least work with the existing one and potentially make a suggestion about what I might do with another department.

That’s where the fourth leg of this framework is, again, thinking about value streams, not just departments, but what we’d advise our clients to do is go ahead and think about automation from a value stream perspective. And if you’re only willing or able to make a change departmentally, do it by all means, but don’t flip that. Don’t start with departmental optimizations blinded.

Ed Haines

Oh, right. Yeah.

Sean Woolley

… to value stream improvement.

Ed Haines

Yeah.

Brian Betkowski

Good.

Ed Haines

Good point.

Brian Betkowski

Well, this is good. Anything else, Sean, that we didn’t cover that you want to talk about or is that good?

Sean Woolley

I think that’s really great. It was a good conversation. We are having so many discussions with mid and C level leaders out in the market on this stuff. It’s certainly a topic of great interest and I’m looking forward to seeing our clients continue to have great successes with this stuff.

Brian Betkowski

Yeah. If any of our listeners urge you to reach out to Sean, if you want any just insights or frameworks or anything else to help you on the journey, be happy to help.

Sean Woolley

Yeah, happy to help.

Brian Betkowski

Thanks for joining us today, Sean.

Ed Haines

Thanks, Sean.

Sean Woolley

My pleasure. Thanks for having me.

Brian Betkowski

Cheers.

Ed Haines

Cheers.