Ganes Kesari: 0:04

A lot of people talk about picking the right solution and they talk about AI, computer vision, and other technologies. Yes, that's important, but the bigger challenge when it comes to data science applications, is in helping users adopt it. So if building applications is tough, helping them adopt, it is even tougher

Tom Raftery: 0:22

Good morning, good afternoon, or good evening, wherever you are in the world. This is the digital supply chain podcast. The number one podcast, focusing on the digitization of supply chain. And I'm your host global vice-president at SAP Tom Raftery. Hi, everyone. Welcome to the digital supply chain podcast. My name is Tom Raftery from SAP. And with me on the show today, I have my special guest Ganes. Ganes, welcome to the podcast. Would you like to introduce yourself?

Ganes Kesari: 0:49

Thanks, Tom. Absolutely. So I'm the co-founder and Chief Decision Scientist at Gramener. I have about 20 years of experience solving organizational challenges using technology. I co-funded Gramener to help organizations make better decisions with data analytics and storytelling. In my current role, I lead our client advisory and innovation, helping CIOs and chief data officers come up with the data strategy, identify strategic initiatives as part of the roadmap and ultimately get business ROI through data driven decision making. So I am passionate about, uh, writing and speaking on, on the applications of data I write for Forbes and Entrepreneur. And I have spoken at events such as, uh, TEDx, uh, Inbound and I also, um, am passionate about teaching. I run guest lectures at schools like Rutgers and Princeton university. Overall, I'm obsessed with data that's the underlying message. I've uh, for example, I have been logging my time 24 hours, every day for the last four years. And I've analyzed this data to build habits and improve my personal decision making.

Tom Raftery: 1:59

Well, that's taken it to a whole another level. Well, okay. Very good. Very good. So you are passionate about data science and data science strategy, and you help organizations with their data science strategy. What kind of challenges do you come across or do the organizations you help come across? Uh, primarily what are the big, main challenges that organizations come across that you can help them with?

Ganes Kesari: 2:27

Yeah. these are making better informed decisions, whether it is a strategic level for executives, when they want to decide where to take the company next, or it could be at the middle or the ground level, operational decision-making. So organizations reach out to Gramener, to. Not just get advisory on how, on, where to go, but also implementation support on how to realize the value. So we do both, uh, likes of data, strategy and roadmap I talked about earlier and we have a low code platform, which we use for implementing the solutions and implementing machine learning and data storytelling so that you can realize the value by connecting data, to decisions.

Tom Raftery: 3:10

Okay. And was it this low code platform that you use to analyze your own data?

Ganes Kesari: 3:16

So this, I have done it on Excel, I believe in, uh, using the simplest tools possible and then scale up in complexity as you need it. So for this one, I'm still at the level of using descriptive and diagnostic analytics uh, with Excel. But then the next level I've been, um, um, now, and I've been doing it for four years, I'm looking to scale it up and automate some of those. That's where, uh, perhaps the next stage a low code platform will become very handy.

Tom Raftery: 3:41

I can imagine because it doesn't Excel have a, a limit to the number or rows in every spreadsheet?

Ganes Kesari: 3:49

No Xcel actually, uh, the, the current version goes beyond that. I think we used to have, a certain limit, but now. I don't think that exists anymore. Uh, and Xcel, by the way is very powerful. It can do a lot of stuff just that when it comes to automating, a macros can take it only to a certain extent. When you want uh, to have end to end automation or even some visuals, which bring out the insights automatically uh, you will go beyond the limits of Excel and you'll have to look at other tools.

Tom Raftery: 4:15

Okay. Okay. Well, we'll come, we'll come back to supply chain. If, if organizations are in the supply chain space, where should they start to transform their operations using data?

Ganes Kesari: 4:31

Yeah. this is a common question. I get, any organization starting with the data. So the common challenges are. They get started with, uh, an interesting project in data and analytics. I often say for success with the data and analytics, start your project, not with data or with analytics, but with business challenges. So for supply chain, again, you'll have to start with the organizational strategy. Where does the organization intend to do, uh, the two to three years timeframe? And what is the priority for the organization this year? And then, identify the data and analytics strategy based on the organizational strategy to see what decision making you need to enable and who are the users you want to support. So that's a good starting point, rather than jumping into executing a few projects,

Tom Raftery: 5:20

Okay. Are there any particular industries that are further ahead in this space or that you think need a bit of a leg up? Or anything in-between?.

Ganes Kesari: 5:31

Organizations or industries that are different industries, for instance uh, financial services is one industry which is ahead. And, uh, there's an interesting report I've come across, from International Institute of Analytics, they have ranked the different industries based on the maturity, the data and analytics maturity. So, so that way, I don't think supply chain falls somewhere in between. Um, there are, uh, leaders like financial and other industries, so overall it will be a useful thing for organizations to see where their industry lies and within their industry. Obviously there's again, a good spread. There are leaders and laggards, so, so that's a way to, to benchmark oneself and see how one compares over time periodically every six months or so.

Tom Raftery: 6:15

And what about the leaders themselves? There has to be some kind of a culture shift, in organizations to adopt a data first strategy. Are the leaders within those organizations, do you think they are, do they have the right culture for doing that? And if not, how to change that culture?

Ganes Kesari: 6:36

Uh, There are some specific characteristics which standard for, the leaders in those, in the different industries. Uh, One, Uh, clear factor is the role of leadership executives and leadership, where executives understand the importance of data and play a hands-on role in leading it from the front, uh, those organizations benefit. And, uh, a second one you've touched up on, culture, overall within the organization, how ready is organization to accept change? So change management is a second factor. If an organization is very conservative, slow to adopt innovations, they will have challenges, um, adopting data driven innovations as well. So that's the second factor. And the third one is how the organizational processes are, wired or rearchitected to facilitate the use of data. Often people view this as one tool, like a, um, uh, like a bolt-on tool. You bring it in and then give it to people and let them get the benefit out of it. It can help you get incremental gains, but that is not the most efficient way. When you bring in a powerful tool, a data science solution, you'll have to rearchitect the business processes. And, um, so that again goes very closely with the change management and the, the role of leadership. When you have all these three aspects coming together, those organizations, um, gain the most and are able to demonstrate business benefits from data science.

Tom Raftery: 7:58

Okay. And if I am an organization realizing that I'm behind the curve in my data science strategy, what are, like first steps I should be taking to catch up again?

Ganes Kesari: 8:12

First step would be once you understand that, yes, there is a lot of work to do get started with framing, a data and analytics strategy. I mean, briefly touched upon this. We talked about how you can get this started by looking at the organizational priorities, onboard the business leadership and come up with, a strategic roadmap in terms of where you want to apply data and analytics short-term to long-term. So that's a first step and second You'll have to translate that into projects. And this is again, a common area where organizations make the mistake of picking one or two projects either based on, whatever they think is important, whatever is urgent and important, or based on the loudest voice floor who is asking for, uh, certain solutions, the loudest, they go with that. That's again, a mistake you'll have to pick a portfolio of projects. In supply chain, for instance, let's take the case of a warehouse. If a rather chief data officer of an organization, which is running warehouses has a budget of say a million dollars. A typical tendency is to split this budget across three or four functions. Say, give one project to marketing, one to operations, one to finance and try to satisfy everyone. That doesn't work out very well. Instead if you pick one or two strategic areas, let us say, operations improvement and you do three or four projects in that area. Understanding your demand and supply. And second, for instance, you streamline the appointments that you handle at the warehouse and, and other aspects that you pick all those, uh, initiatives, which will give you the biggest business benefit and which has some synergy with other, projects that you're choosing. So that's what I mean by a portfolio of projects. When you choose a portfolio of projects, uh, which is aligned with the organizational strategy and data analytics strategy that can, move the organization, in terms of, the direction and in terms of the business benefits that they can get. That is very powerful.

Tom Raftery: 10:10

Okay. And what kind of benefits then can organizations get? Have you got use cases you can speak to?

Ganes Kesari: 10:17

Yes, absolutely. So let me talk about, uh, one of our clients, a leading cold chain warehouses, now operator, uh, in US. So the challenge they were facing, was to streamline operations. And, one of the first problems we, we picked and helped them solve was, handling the, the turnaround time delays from the time. So, uh, in this case, let's take a truck coming into the warehouse and, the the. time plan to handle this. Uh, the inbound is about the SLA of one and a half hours. And, uh, there were several appointments which have taken more than two hours. How can data and analytics help for that? We looked at several factors. What are the factors which, which, uh, lead to the delay. It could be that you're over booked appointments, or you've taken too much of time on one appointment that has a cascading effect. Or third, there's an incoming truck, which is delayed, which can again, have a ripple effect on all other appointments. So there could be several factors which lead to delays, and, and can have, impact through the entire day. We built a machine learning solution by analyzing the factors, looking at, which are the most important ones. And the solution suggested an appointment time a when a carrier calls. They say that this is the, say you have to come in at 3:30 PM. And fire, that, uh, factors in a lot of these other aspects of which for instance, um, how many appointments do they have and what are the kinds of appointments? And what's the typical delays they have seen for these kinds of appointments from these carriers by factoring all of these, it suggests an appointment slot, and that helps to reduce the overall turnaround time. What we've seen is that, there was a 15% improvement in turnaround times after the solution went live. This was done as a pilot in a couple of warehouses. Eventually it was rolled out across dozens of warehouses, for this player. And there was a, a business benefit projected benefit of a $1.2 million annually. Just by this one solution, to intelligently pick appointments and improve the turnaround time.

Tom Raftery: 12:21

Wow, impressive. Any the other good ones you can talk to?

Ganes Kesari: 12:25

Apart from that, another common challenge today is, the worker shortage and appointment, uh, in terms of how do you plan the work activities within warehouses? So this was also a challenge for our client. So once you have this appointment, I talk about synergy between the projects, right? Once we had the first project implemented a natural second one was to address this challenge of, uh, streamlining the worker productivity and improving that. So here we built a system to identify when is the right time to start planning, to serve as that appointment, because then you have a truck coming in. Now, in this case, let's say there's an outbound. You have to load goods onto a truck which comes in. So you'll have to start picking all the goods and assembling them in the warehouse so that once the truck comes in, you're able to load it onto the truck, uh, in the shortest time possible. So when do you start selecting the goods from different parts of the warehouse and assembling it? That's one challenge. And second, how do you assign the work in a most efficient manner? It's done by supervisors, but when you have so many jobs and different complexities and different types of jobs, can you have a system assist the supervisors? So that's what we did with the, with the worker allocation and task planning system. That looks at a lot of historical factors, how much time it took to pick a certain type of task, like a bulk bulk picking. and, uh, what does the efficiency of certain workers and for certain types of tasks, we're looking at all of these factors.

The system says, if you have a 7: 13:54

00 PM appointment, you'll have to start

at say 10: 13:58

30 AM, or maybe you'll have to start the previous evening. So that's, uh, that's the, it comes up with the, uh, kind of a project plan automatically. And second, these tasks if you have these 15 tasks, you'll have to assign these tasks to these people. And, um, it goes down to a minute by minute level that, okay, this person is currently in this part of the warehouse. So the next task can be given to this person because this person is already there. So that way, optimizing that even, within minutes and, and allocating the work. So when you have all of these coming together, you can imagine the complexity which has gone into building the solution, but overall for the supervisor, they don't have to understand the complexity of data and analytics. They just have the system making recommendations and they can choose to follow it or choose to override it in certain exceptional cases. So that again is a solution which we built, which has, has just gone live, and you're seeing very good benefits there as well.

Tom Raftery: 14:52

Okay. And where to from here. I mean, what's, what's next for data science. What's the next big thing coming down the road?.

Ganes Kesari: 14:59

Yep. Yep So if you look at, uh, warehouses, um, in the future, you'll have, and we are seeing that, but some of the organizations like say Amazon and others who are some of these digital native organizations who have implemented a variety of, uh, automation technologies. Uh, so you have, uh, robots and others within the warehouse. So in the future, we will have a warehouse, which, completely is automated. And whenever you have these, the appointments being booked by, by customers. So you have humans coming in and intervening in certain parts of the activities, but, but for that, everything else, is mostly automated. So I imagine this like a black box warehouse where you have everything happening in synch automatically. And you just need to feed in the inputs in terms of, Hey, I need have these many customers coming in today and these many customers are sending the trucks to pick up the goods. So if you have all of those, business inputs provided to the system, it automatically manages it. And, does things beautifully and in a repeatable manner. So that probably is a future where we are heading to.

Tom Raftery: 16:05

Wow, fascinating Ganes we're coming towards the end of the podcast now, is there any question I haven't asked that you wish I had or any aspect of this that we've not discussed, that you think it's important for people to. be aware of?

Ganes Kesari: 16:19

Yep. One aspect I briefly touched upon. Um, I think it's worth emphasizing is. Once you have a solution, a lot of people talk about picking the right solution and they talk about AI, computer vision, and other technologies. Yes, that's important, but the bigger challenge when it comes to data science applications, is in helping users adopt it. So if building applications is tough, helping them adopt, it is even tougher. This again needs, factors like we talked about the role of leadership, rearchitecting, the business processes and change management, but it needs continuous attention throughout the solution design phase. Again, to give an example from, the work at Gramener, we've seen the best results come when you're involved with the business teams right from the time you choose the projects so that there is a sense of ownership and they get involved, when you're building the solution and, um, on a weekly or fortnightly, because they get involved, they review the solution as it is being built. And that way they can act as ambassadors to take the solution to other parts of the organization, rather than our technology team, taking the ownership and trying to educate and train the users. If you have say for instance, warehouse managers or the business leaders playing that role that is seamless and very effective. So that's again, important for adoption. You'll have to start, right, but the right strategy and altogether to focus on the last mile and ensure options, right. That's very important.

Tom Raftery: 17:47

Fascinating, Ganes, that's been really interesting. If people want to know more about yourself, or about Gramener, or data science or any of the things we discussed on the podcast today. Where would you have me direct them?

Ganes Kesari: 18:01

Yeah, Gramener the website. We'll share the links in the podcast at the end. So the website of the get a good place and we are active on LinkedIn. And, um, I'm personally, I'm active on LinkedIn and Twitter as well. And I have a newsletter I publish on a fortnightly basis. These are good points to get in touch.

Tom Raftery: 18:19

Cool. I'll have links to all those in the podcast notes. Super Ganes has been really great. Thanks coming on the podcast today.

Ganes Kesari: 18:26

Thanks for having me, Tom, it's a pleasure talking to you.

Tom Raftery: 18:29

Okay, we've come to the end of the show. Thanks everyone for listening. If you'd like to know more about digital supply chains, head on over to sap.com/digital supply chain, or, or simply drop me an email to Tom dot Raftery @sap.com. If you like the show, please, don't forget to subscribe to it in your podcast application at choice to get new episodes, as soon as they are published. Also, please don't forget to rate and review the podcast. It really does help new people to find the show. Thanks, catch you all next time.