Find out how RB uses cognitive technologies in processes such as forecasting and pricing and promotions.
How can a large organization use cognitive automation to innovate crucial business processes? That question faced Saqib Mehmood, Global SVP Business Solutions & Insights, with the Hygiene Home team at RB.
For RB, whose products include Lysol and Mucinex, one of the answers was moving away from proprietary IT technology and partnering with companies such as Aera Technology.
Mehmood adopted cognitive technology to automate processes like data gathering and analysis, pricing and promotions, and forecasting to make better decisions in positioning RB products to stand out in a very crowded market. Cognitive automation and artificial intelligence help Mehmood and team crunch staggering reams of data from every part of the business to accurate forecasts of what the market wants and what will help their product catch a consumer's eye.
Watch the video to learn how RB uses cognitive technologies and the cloud to gain a competitive advantage.
The transcript below has been lightly edited for clarity and length.
How large is RB (Reckitt Benckiser Group PLC)?
Saqib Mehmood: We operate in about 60-plus countries. We're manufacturing in a lot of strategic locations. We operate with closer to about, I think, more than 15 global brands.
Why did you adopt cloud-based solutions?
As the IT roles are changing, you're required to squeeze more out of your infrastructure so that you can spend more on innovation. Why is it that I can buy 200 gigabytes of space in a click of a button, as a consumer, but I still spend 6 months trying to do that when I am trying to do it with internal IT processes?
Michael Krigsman: How does that affect your internal IT operation?
We today run our core BI for all the sales, marketing, and most of the financial performance systems which are run on the cloud. When we started the discussion, everyone was laughing us to say they're going to take our core BI with gigabytes and terabytes of data every day being moved around, and this is now a norm. We have taken that journey eight years ago, which seemed a bit out of the box but, at this point in time, you see more of the companies operating their BI on the cloud.
I think these are some of the trends which will follow. If we are trying to build all of these things internally, like local BI systems, et cetera, we will never be able to get the right resourcing which comes with it and, at the same time, I think technical resources, high quality of technical resources will really be competing with all these big companies who are operating in this space. We would rather like to operate in a space where you get great quality of resources which are business-facing, front-line focused, trying to solve problems or businesses rather than building technology.
What is the business impact of cloud at RB Group?
You have to have your business also engaged on this journey because, the fact of the matter is, it is 90% change management and 10% technology when you want to go on a journey like this.
Michael Krigsman: Tell us about change management?
We made a massive change, even the way that I wanted the role to be positioned. We changed the name of IT to business solutions and insights. It wasn't about just delivering technology. It was first delivering the right business solutions followed by insights, which are then powered by the right technology platforms.
It took a lot of time for the business teams to understand that we are not order-takers. We've become more order shapers so, when someone tells me, "I need something," the first question you ask is, "What problem are you trying to solve? What are you really trying to achieve out of this?" That's a very different conversation for a lot of the business teams. Every CEO will tell you that he wants the IT organization to be a value driver but that requires a completely different mindset of how you are structured, how you operate.
Last year, we made a big play. We made a big play to say we will rename ourselves. We made a big play in terms of saying that 80% of our time will really be spent on transformation agenda rather than the running agenda, and we will start to use the cloud as one of our biggest enablers to take the 80% of focus on actually running the business to 20% and start moving a lot of the focus there to move these things on the cloud.
How did you move to the cloud?
It's easier said than done because that requires a massive change in your organization. It requires a completely different mindset. Then, for you to be able to prove to the business that you can actually play in this space because this is not a space which is given to you. This is a space which you earn, and it is only earned based on you being able to drive the business in that direction.
So what we did a couple of years ago when we started to frame the whole agenda is, like we said, we didn't start by saying we want to hire a data scientist or we want to take over everything that is done in analytics. We started with a very clear agenda to say what problems exist in the business today that, if I can solve, I can help you to transform.
It's actually interesting when you ask that question to the business. It's not an easy one for them to answer either. You ask the supply chain team, "What can I help you to solve?" You come back with a list of 600 things that you can do.
What you realize is, most of it is not really problems; it's just symptoms. Then to take the organization through to say, "Okay, you've given me 600. Why do you need them? What really helps you to transform?" To come out with six, in our case.
We came out with six transformational agendas which we thought would help in terms of transforming our function as well as our company, which was in the area of supply chain, which was driving more predictability in our supply chain, which was in the area of sales on how you could actually make a sales employee's life better by giving them all the stuff that they do today through manual computations, et cetera. Making their decision-making power more enhanced.
We started talking about, with our consumer teams, how we can use data as an insight into the innovation process in the whole marketing, targeting, and hyper-marketing process. It required us to be very sharp by taking 800, 900 of those things and then making them into very concrete areas which are digestible but, at the same time, exciting enough for someone to say that if you really intervene in the space, wow, you can help me to transform.
How does cognitive automation help with pricing and promotions at RB?
We are an FMCG company. We operate through the Walmarts and Targets of the world. For us to be on that shelf clutter, for us to be noticed, and for consumers to pick us up, you have to have the right pricing, the right promotion, the right package.
You get the four Ps, right? The four Ps really play. Once you've got a superior product, you have to win on shelf with all the clutter which is on there. That's the biggest challenge that a CPG company has. How do you become relevant on the shelf that, when a consumer is passing by, they pick you up?
Our teams, of course, then spend a lower amount of time making sure that it's not just the packaging or the superior product that we have, but what price point do we put it at; what's the promotion that will work? It's easier said than done. We've got thousands of SKUs. When you're negotiating with the buyer, you're working on the category level. There's sometimes when the buyer is not agreeing to some stuff because he wants to make a margin on it. It's a very complex world.
Michael Krigsman: How much data are we talking about and the types of data sources that you look at?
There are many. You have your own internal sales data, which I think is a starting point. You have the data coming from what we call the scan or point-of-sale data which comes from our retailers. We use that as an input.
Nielsen is providing a lot of information for us also in terms of what's working just not for us but, also, for the competition, pricing points, et cetera. We also use a lot of the Amazon data because we do have a significant business online to see how we are performing versus competition ourselves, what pricing points work, what coupons work. Then, of course, our supply and inventory data. All of this is terabytes and terabytes of complex data which, if you were to give it to a normal individual and say, "Now you use all of this and try to find your way out," good luck.
How do you manage large volumes of data for cognitive automation?
We started engaging with a company called Aera Technology. Aera came with a very exciting proposition for us because usually what happens with most of the companies when you start to work with them is that the system which is around are really focused on transactions. I think if you look around SAP, most of the systems which are there for sales force automation, for promotion management, they're really focused on the process.
This is how systems were, in the past, designed. They were designed for process compliance, not for being smart. I think that is one area where we started working with a company called Aera, which doesn't make these systems redundant but starts to come on top of them and starts to make these systems smarter.
If you are able to take all this transaction data, all these gigabytes of data that we talked about in different systems; if you are able to pull this through; if you are able to create a network around these different data sets and make some meaning out of them, and then, on top of this; if you are able to use data science and smart algorithms which are able to do this computation which is not possible for a human being going through 36,000 different combinations that can get you the best pricing point that you feel that will make you win; is what Aera does for us.
Aera takes all these disparate pieces of data which we have, bring it together, put some logic around it to clean it, put some algorithms that work on top of this, and then it becomes a far smarter output that can be used for our key account managers to start taking some decisions rather than trying to try to figure out what data works and what doesn't.
How is cognitive automation different from traditional computing?
Because you have all these systems which have their own data sets, they all have their own master data. They have their own connections. Trying to put all of this in one place has always been a mess. It's not always easy to take 15 systems which have their own master data definitions and then trying to link them up as one holistic view.
I think this is where Aera really helps us is to create a cognitive data map. Once you've been able to do a cognitive data map, you are able to look at data all the way from your storefront all the way to the supply chain. This is a huge power which you can't do when you have data lying in 15 different systems.
Then, on top of this, adding smarter algorithms that can go across this whole piece off of data maps which are there and then being able to find for you the nuggets of information is not just pure competition because I think it's also, apart from computation, about the machine learning that comes with it. It learns every time. As you as users start to accept something, it starts to make it more and more powerful in the system in terms of maps. If it starts to see that users are rejecting some recommendation, it tries to find out why and take them through the elimination process.
I think the biggest benefit which I see is not about computation. It's about creating a brain within the company which used to be all in employees and then, when someone leaves, it leaves with all that proprietary information. It takes you three years to get a key account manager really understanding some of the staff, how these things work, and then, when they leave, they leave with all this data and knowledge with them.
I think being able to preserve the different data elements, how each promotion work, trying to make a map out of this, and then being able to preserve it every time someone makes a decision on top of it, is actually super powerful because the new guy who comes in into the job can pick it up from where the last guy left it. For us, it's not about just being smarter decisions. It's about retaining the knowledge within the company as well.
Do data, algorithms, and the maps serve corporate knowledge retention?
Exactly. Exactly, and that's why it becomes a stronger proposition is, how do you retain brain power within the business and really make this as one of your critical brains? It should work much smarter than any other CAM and then the CAMs can focus more on relationships. They can focus on more negotiation power rather than spending hours trying to compute what is the right pricing point or what's the promotion I should be running. They should be more focused on how they can drive business from a top-line perspective using these tactics rather figuring out what tactics work and what doesn't work.
Like we said, this is change management because the first time you give someone some results, they say, "Well, I can always do this better than you because I don't trust the machine doing this for me." I think it's change management.
Ten percent is really what I talked about. It's very powerful, but this is 10% in an organization; 90% is change management. It's about sitting down with these people, taking them through this process, making them feel on why their job can become far more strategic and important rather than trying to crunch some of this stuff, which we can do for you in a far, far better way.
The analogy I give most of the time to my teams is, you never get up in the morning and you say I'm going to forecast the weather. You have weather forecasting systems that do the weather forecast. You plan around it. If it's raining, you take an umbrella.
This what the function of the business should be. They should not be doing things which can be done better, faster, efficient in terms of predictability, like forecast, like promotions that were wrong. They should plan around them.
Once you know what's going to happen, you plan your tactics around it. It's a far better forward-looking way of operating a business than trying to spend a lot of time predicting what machines can do faster, better.
How can cognitive computing and automation help forecasting?
For forecasting, also we use the same company, which is called Aera. Before we talk about forecasting, it's not about forecasting. It's about bringing predictability in the business because guess what happens. One wrong decision, which is made five weeks ago, can have multiple ripple effects when it's actually in action.
If a customer places an order today, had I been able to predict that much better five weeks ago, I would have been able to get the raw materials in time, get it produced on time, and shipped to the warehouse. That time dimension is actually the value. The value, that's what changes companies. The lesser that time becomes, the more challenging it is.
If you have three days to actually do it when an order has come in, you actually start to create a huge ripple effect in your company. You start a panic by which you start going back and then you start changing things in your supply chain, in your manufacturing, and then it just becomes a hassle. The more people who touch this at the last moment in time, it just becomes a continuous web of a lot of mistakes that the lower end of the stream is handling, which is really your supply chain.
By just bringing more predictability in your value chain, you can eliminate a lot of these things. That's where we started from. What is the purpose that we're really trying to solve? We wanted to make sure that we can, with five weeks in advance, have better predictability on our products and then use this as a driving point for what really needed to be changed at a macro level.
If we knew that a customer operates in a certain way, and the second week of the month I will get maybe an unplanned order because this happened three months ago, and the three months ago because he does some inventory changes in his system and, hence, some events get triggered by this, I can use this now to start planning much better.
What are the data sources for trade promotions?
Ninety percent of the data sources are the same which you use for promotions because, at the end of the day, sales data and promotion data, most of them use the same. You have financials and inventory. You have what's your raw materials in the factory. You're taking most of the data coming out of your ERP. Then you take the scan. We use a lot of data from Nielsen. We use a lot of data which is coming through third-party systems like merchandising, OSA, which we feed in.
All of this comes in, in a similar way. We try to make sense out of it, create a cognitive map, and then we let the system run some smart algorithms to start doing the prediction. You have to train because, like we were discussing for the trade promotions, the first time you would give a promotion out, forecasting out to the planners, they would never accept it. But then, you start saying, "Okay, I'm going to start competing with you, so you do a forecast and we'll let the system do the forecast. Then let's run it for three months, and let's see where are we getting better."
Then you start getting on the journey because, after some time, they start to realize that actually all this time that they spent, they could have done something better. As you go through this, I think that's the most important point. You have to take people on this journey rather than trying to make them compete versus the system. This is where most of these change management programs fail is when you take a lot of these nice, fancy gadgets and try to threaten the employees that they might not have a job.
Does this free up employees' time for innovation?
For them to be standing in front of their customer teams and being able to show that they are actually doing much better in terms of forecasting, hence service is a far more encouraging thing for them than to say that I missed out on some things. Since they have 80% more time because we've really now reached a stage where half the business in the U.S. is run without anyone touching it, so it's really the machine that decides what we predict, these teams have more time in terms of taking these peaks and troughs and trying to normalize them, working with customer teams, working with the manufacturing teams, and making sure that they can use their time for more planning rather than forecasting.
What advice do you have for Chief Information Officers?
I think it starts with purpose. What problem are you really trying to solve? Most of the time you just starting, walking on a journey without trying to understand. We are not a tech company. We are an FMCG company. We sell brands and our problem starts with brands, with customers. When you put this at the core of your problem and trying to really solve problems for customers, for consumers, and for your brands, then it's a very different conversation.
Also, trying to make sure that you are not an order taker within the business. You might be appreciated short-term for just taking an order and then executing it, but this is not where a value-added, long-term role, relationship, or respect can develop.
Initially, it means you have to grow a bit tough. A lot of people will come to you by putting 50, 100 things into your transformation roadmap. Making sure that you can deliver, having tough conversations to really find out what will move the needle and what really drives company purpose and the results are, I think, the starting point.
Published Date: Jun 26, 2019
Author: Michael Krigsman
Episode ID: 605