Tom Stephenson:
Good morning, good afternoon, good evening, wherever you are. Thank you again for joining us for the Cognitive Automation Summit. We're very excited to have you with us. And, we're also very excited to have Suketu Gandhi from Kearney with us. Suketu is a partner at Kearney, and he's responsible there for the global digital supply chain practice. And in a part of that practice and part of his work with clients, he serves a variety of consumer and retail clients. Suketu, welcome aboard.
Suketu Gandhi:
Hi Tom. Wonderful talking to you.
Tom Stephenson:
Thanks again for joining us. Today, we're here to talk about sense and pivot, which is an interesting area of research and thought leadership that Kearney's been at the forefront of. Suketu, do you want to tell the folks that are listening in, exactly what you're seeing? What's the research that you've done around this opportunity?
Suketu Gandhi:
Absolutely, Tom. This topic really surfaced for us about a year and a half, two years ago. When we were seeing that supply chains that were built for an era, where what happened yesterday will happen again tomorrow, were not working out too well. We were seeing dramatic increases in the capital needed to operate supply chains. The second thing was out of stocks were going up, new product introductions were struggling. And, what that led to was an epiphany for us at least, that said that if you wanted to design a supply chain, then you really had to think about a model that allowed you to operate in a way that said, "If something happened today within my business, as well as other parts of the business. If somehow the customer changed, the competitor changed, the customer, consumer changed, how do I react to it?" And what that led to was our thinking around blending digital and physical together, bringing artificial and human intelligence together.
Suketu Gandhi:
And finally last but not the least, have the ability to constantly track and change, which is what sense and pivot is. An easy way to think about it is Waze. All of us use Google maps to start with. And now, with Waze you get real-time data. It doesn't take the human being out of the decision-making process, but it enables the human being into making better decisions. So, that's where sense and pivot started. And the whole goal was to ensure a supply chain that is serving the consumer and customer significantly better at a lower cost. The one new element Tom that has come in, that I would like to add is in COVID. COVID has completely disrupted supply chains. And, clients that we have worked with in sense and pivot were able to respond significantly better. And now, the topic or the word du jour is resiliency. And again, that's an area in which we've done a lot of research and that has helped our clients deliver their promise for their consumer.
Tom Stephenson:
So, what has happened to those companies that you've seen, that if I follow your map-making analogy with GPS and Waze, maybe even go back to paper maps, the people that developed and run their businesses on predictions of a monthly plan or even a quarterly plan. What has happened to those companies during the COVID time period?
Suketu Gandhi:
The classic Mike Tyson quote that, "Everybody has a plan until they get punched in the face." We're allowed to say that, nobody dies on Mount Everest because they didn't have a plan. The reason things fail is because, one, you're not constantly listening to what is happening. The second thing is you're not reacting or don't have the ability to react. And I think, the ability to react is a critical component. So, folks who are seeing dramatic changes in demand or supply, is putting their value at risk at a significantly higher number than it was before. And now, they have succeeded, in some cases, through sheer force of will. Either, they have reduced the number of skews they stock. They have changed the way they deliver. They've had to change the way they package. And then finally, it has all had an impact on the amount of choice you can offer your consumers, your customers. So, folks who didn't have this in place reacted, but in a manner which wasn't most optimal for the consumer or the company.
Tom Stephenson:
They were in survive mode, versus thrive mode.
Suketu Gandhi:
Well put, well put.
Tom Stephenson:
And so, let's talk about sense and pivot in its pieces here. So, what does it mean from a sensing perspective? What information are you looking at? What's the frequency? Give us a little bit of a feeling for what you've seen, in terms of what's world-class sensing?
Suketu Gandhi:
Absolutely. Let me first lay out the story for sense and pivot, right? What are the three big components? The first component is sensing. And, that is about looking at your internal orders. What happened yesterday? Seasonality, things of that nature. Second, is looking at your consumers, how they are behaving, what are they looking for? What is happening in the market itself? The third component is what is happening on the supply side? Are there any disruptions coming through?
Suketu Gandhi:
And putting it all together, you get a sense of what is truly happening in the marketplace. The second part is a cognitive engine, which Aera has a wonderful tool in place that allows you to take all of these sensed ideas and say, "Okay, which of these are signals, which of these are noise? And how do I react to it? What do I do? Do I change my planning horizon? Do I change the way I source? Do I change the country where I source? Do I make it in a different geography, or do I rejigger my transportation network? Those are the answers that come out of a cognitive platform that has deep human and artificial intelligence built out.
Suketu Gandhi:
And then the third component, which is the physical stuff, "How should I organize my people to make better decisions? What do I do with my DCs? What do I do with my retail stores? What do I do with my manufacturing facilities?" So all these three components together, make up sense and pivot. But, to your core question, what is world-class sensing? World-class sensing is truly looking at everything that is internal or external that can tell you anything about what a consumer or customer might look for, and what are the hindrances today in you delivering to them, and to that promise. That's what world-class looks like.
Tom Stephenson:
If I think across those three sources of information, the internal transaction data, the information about consumers, and the information about supply, is the information in the enterprise today of a high enough quality today to be able to provide a good signal into sensing?
Suketu Gandhi:
So, this is classic, right? Because, everybody says I don't have the data. And still, we all make decisions, right? At the end of the day, human beings are decision machines and we are able to make decisions. But, the critical thing is we are going from a deterministic thinking that says, "This is right or wrong." To statistical thinking that says, "I have 90% confidence plus or minus 5% on what could happen." So, data is going to be dirty by definition. If data was clean, everybody would be making the perfect decision and we'd be in a world where we'd say, "Everything is rational." We know we are predictably irrational. So, what you do is you take the data, you put a degree of confidence on the data, as well as the decision that is coming out. And then, you make a call as a human being, "Am I confident with, or am I not?"
One of the powerful things about using AI for this is to say, "Hey, out of the 80,000, 100,000 SKUs you have, we have a high degree of confidence on 80,000, the other 20, we don't know. We have low confidence." So, that's where the human being comes in. And then, the last component is what the model should learn. So, it's again, not an on-off switch, but a dimmer switch, that you get better as time goes on. And, the machine teaches the human and the human teaches the machine.
Tom Stephenson:
So, you talked to a bunch of clients out in those sectors of consumer goods and retail, what are they looking for today? Back to your second point of your journey on sense and pivot, around cognitive and the use of AI, where are they in their journey, how are they thinking about what they're looking for?
Suketu Gandhi:
To every single person, especially the leaders in the world. In the last six months, I've probably talked to 200 supply chain professionals across Fortune 1000. Nobody has said that this idea is not valid. They've all been thinking about it. What they're saying is, "How do I make it happen in my organization?" First question is, "What do I do with the data?" Second question is, "How do I build my skillset to make these decisions?" And then, the third question is, "What is the tool set I use to deliver against that?"
Then they go deeper into it and say, "Okay, that's wonderful. Agreed, you have a plan. How do I break it up into small chunks, so that the change process of my commercial, my finance, and my supply chain coming together can agree upon a path forward, so that you deliver small things. Do I start with a particular product, or do I start with a geography? Do I start with a channel, right? Whether it's physical or e-commerce, or third party. How do I make those decisions? And how do I make this real?" Nobody disagrees with the fact that digital supply chain and using data to drive supply chain is given, but they are all asking, "How do I make it real?"
Tom Stephenson:
Yeah, it's interesting. It's two dimensions. It's where to start, those are the dimensions that you were talking about was, criteria on how to define the pilot area, because they want to make sure that it works before they scale it up broadly. And, on the data and the technology side, they have lots of choices. So, it's difficult to understand how to put that together. Particularly in some cases, where they may feel like they have to go to lots of different partners to get it.
Suketu Gandhi:
Yes.
Tom Stephenson:
How are you advising them on those two dimensions of where to get started? How to think about the scope and how to think about what pieces of technology?
Suketu Gandhi:
Hopefully, this is the only consulting phrase that I'll use, but it is, think big, start small, and scale appropriately. And, let me break it down. This model of let a thousand flowers bloom, and have many use cases running around a whole company, when you are at 20, 50, 100, $500 billion company, it doesn't work. You really have to have an end goal, a simple star where you want to be. And then, work backwards from that, because that determines what are the metrics you're really going to affect, right? Are you going to increase your customer [inaudible 00:12:42], on much better NPS score? Are you going to have a higher gross margin or top line? Are you going to require lower capital to drive this business? And a lot of times, it's a combination of all.
So then, get the metric right, understand the end goal, and then drive these pilots towards the land of scaling, because for a $50 billion company to move the needle, it's got to have a significant payoff. Otherwise, it's a series of interesting experiments. Now, the one wonderful thing about this crisis we have right now is CEOs have taken ownership of this topic. Supply chain is being talked about at the boardroom as a topside topic consistently. And what that has done is it has changed the focus from interesting pilots to scale. And to scale, it's back to thinking big, laying out the architecture, understanding the operating model, how will you organize your people in your physical assets, and what will you do with your technology? And, even if it's not perfect, how do you get to the payoff that you're looking for? And, that's what folks are really working through today.
Tom Stephenson:
Let's talk about those two topics that you just laid out, operating model and technology. So, let's talk about operating models first. So, a company believes that it needs to move to a different way of sensing what's going to happen, and reacting to that in a very agile way. What did they have to do to change the way they work to make that possible?
Suketu Gandhi:
Yeah. So, the biggest thing is, it starts with the mindset. How do we change our mindset from thinking that I'm going to have an organization that is going to sit there and worry about, for example, planning all day. And then, plan for a year, quarter, month, as you talked about. This is about almost real time, if you can make those decisions. Changing your time horizon is really critical. The second way you change it is the granularity of those decisions. Gone are the days where you could say, "Oh, I'm going to think about a product family." And, plan according to that. This is at the skew level. The bag goes out at the individual consumer level, because your competition is thinking about consumer skew, and today. So, how do I adjust to that? And then, how do I organize myself, so that decisions are made significantly closer to the geography product, at the same time as something in the central area that can say, "Hey, these are the broader trends we are seeing." So, that's the people aspect of it.
The second aspect is, expecting failure. Now that's a hard one, that means that the way my DC operates will break, if there is a significant change in demand, and that is about my threshold. So, what do I do? Well plan for it right now. That's the heart of resilience. You might say, "75% of my DC is laid out according to how the products are there. And, 25% around how orders come in." That changes the game. "What do I do with my manufacturing platform? Do I bring flexibility there, or have some third party manufacturing locations that can bring me additional capacity to scale up and down? And then, finally last but not least, is your sourcing and network. How is that going to come into play? But thinking about all of those factors together, and that's the key word, together, not as standalone. Because we've seen a lot of examples of fixing one, but causing a problem somewhere else. So you're fixing sourcing and suddenly planning fails because now you've made a decision that is actually counterproductive for your resilience in the long term.
Tom Stephenson:
As we'd say in the United States, "The whack-a-mole problem." You hit one place and it causes a reaction someplace else.
Suketu Gandhi:
Absolutely right, Tom.
Tom Stephenson:
So, on the technology side, I've heard you say a number of things through the discussion so far. You've talked about the breadth of data, the need to have as clean a data as possible, recognizing that data is inherently dirty. You've talked about the need to get to a very frequent level of refresh. You've talked about the need to be quite, quite granular down the skew and location. And you've talked about scenario planning and scenario modeling that enables that. How do you think about what technologies to employ? How do you think about how to leverage what clients already have? Talk to us a little bit about the technology.
Suketu Gandhi:
On the technology I've typically broken it up into three components. One is the classic good old transactional technology, things that accept an order, things that decide what your reorder levels should be, once it's set, things that allow you to understand when product might come in or sub-components might come in, or suppliers might stay in your stuff. That is classic transactional stuff. Every large company that we have worked with has put in tens, if not hundreds of millions of dollars down. Fantastic. Don't try to pollute that system by modifying to work for sense and pivot, leave it alone. The next layer above that, is the data components and data layers, where we bring in data from transactional systems, external third parties, structured and unstructured. Now, that's where a lot of the modern data architectures work incredibly well, right? Whether you use a graph database to go a little bit technical on you, or any other storage device.
Now you have the ability to bring large amounts of information in one common place on which decisions can be made. And then, the third layer is the cognitive layer or the AI layer, that says, "On the basis of all the information I'm getting from transactional systems, from external systems, what signals do I see? What is the noise I see? How do I separate the signal from the noise, and create the right decisions for people to track?" The last but not the least, but this is a critical area, is tracking decisions. So, today you said, "Tom, hey I'm going to order five widgets" of some kind. And then, tomorrow you go and see the same location and the five are still sitting, or you're suddenly down to zero. So, you got to know that the decision that I made was either right or wrong and how do I learn and adjust from that? And that is an absolutely critical component of building this out.
Finally, the third layer also has the component of simulation. The "What if...?" People really have a hard time projecting the future, because they are always tied to what happened yesterday.
Tom Stephenson:
Yeah.
Suketu Gandhi:
And, that is a hard thing to break, but that is another important component of that third cognitive layer that you guys talk about quite a bit.
Tom Stephenson:
Yes. Interesting. You talked at the beginning about demand data, the consumer and supply data. In this new world, there's a fourth data, which is the actual response, "What happened?" You can create a new data set to create that reinforcement learning loop that you talked about.
Suketu Gandhi:
Absolutely. That's a nice way of putting it, Tom. That's well said.
Tom Stephenson:
So, Aera and Kearney have talked about working together. Can you tell us a little bit about what Kearney has seen in the opportunity for our clients?
Suketu Gandhi:
The simple truth of the matter is Tom, in our interactions with Aera, the one thing that really matters to us is focused on an outcome for the client. Fortunately, or unfortunately you guys have the same focus.
Tom Stephenson:
Yes.
Suketu Gandhi:
And so, that's a big thing. That's number one. The second thing we have seen is you spent a decade tackling the hardest problem, which is getting a bunch of messy, crazy data into a model that is manageable. In the old wild west, the guy who was paid the most was a wrangler.
Tom Stephenson:
We're data wranglers.
Suketu Gandhi:
That's right. And, you guys are really good at data wrangling and putting it in a way that it can be used better. And then, the third element, which is built-in on both our sides is the ability to learn from either your mistakes or success. And, that's what today's environment of AI and cognitive science is telling us, can you explain AI and how it made the decision? And, can you learn from it? Because otherwise, these things can become... If somebody called it weapons of mass destruction. And, we have seen that in the past, right? Where people have gone and made decisions and then never gone and then corrected them, right? So, that's what has allowed us to work very well. We have some really tremendous industry and operations expertise. You folks have great cognitive data wrangling, and a broad view of the world from a technology perspective. So, it makes for a wonderful complimentary offering that allows us, again, to serve the client. Because at the end of the day, that's what matters.
Tom Stephenson:
We're as excited about it as you all are. And, the last element that I would add to your description is, in our platform, we enable the capturing of the results of decision, so. Which is super powerful because it allows you to help the solution get better, and better, and better over time. We couldn't be more excited about working with you all to help our clients be successful. Any final words to the audience?
Suketu Gandhi:
The biggest thing is, in the next three to five years, every supply chain in the world is going to get redone, driven by technology, political environment, consumers, and competition. So, it's better to start the journey now, or be left behind, because being left behind would be a terrible place when the world continues to evolve at this dramatic pace.
Tom Stephenson:
Thank you. We're very excited about the story of resiliency and agility that you've shared with us today, based on your research and some of your thought leadership ideas. Thank you very much for joining us, and thank you everybody out there for listening to us. Have a good day, have a good evening. And thanks again for joining us for the Cognitive Automation Summit.
Get social with us!
Follow us on LinkedIn: Cognitive Automation Community
Follow us on Twitter: CognitiveAutomation