End-to-End Visibility into Credit Risk

Gunther Smets:

Good morning everybody. So my name is Gunther Smets and I work with Cargill. So the intent of the next. How much do we have? We have 40 minutes I think. Yeah. So of the next 40 minutes has to go a bit through the technology challenges and the technology design that we take for a credit to cash within the Cargill environment. So not only to the nice stories about it but also a little bit challenging and the dirty stories about how technology is an enabler. But sometimes those who will block or challenge versus what we do. So without further ado quick introduction about Cargill we’re an agricultural industrial business. We exist for over one hundred and fifty years. We’re a family-owned business.

One hundred and fifty thousand employees over 70 business and we’re located in over in the meantime a bit over 70 countries in the meantime as well so we have a global presence. Now we’ve started this journey about three years ago three and a half years ago as a very decentralized company. So every single of our businesses was organized in a completely different way. They had their technology design that had their processes that their organization. So everything was like a single company in itself. So moving from a decentralized with the centralized organization is the journey that we’re on.

We started by focusing on the people so the organizational aspect and really working to a shared service model. So we have a captive share to have a social model with six global centers, so for credit cash that means that we have around 800 people sitting in 600 in the six centers. So that’s 850 FTEs spread over to the six centers that we have to have on a global basis. And that we focused on the process standardization around the process design the presentation here is more going to be focused on the technology element of it as well. But if anybody would be interested afterward to learn a bit more about the process side or the people side on the show to this site don’t hesitate to reach out afterward as well.

So now to quickly recap, so where we started from was more than 60 different organizational models that we had in Chicago. We had over 400 different process variations of different ways of doing things for every single thing. How we did collections how we did cash applications disputes credit management. A lot of our customers are shared across those businesses. So we feel we would establish a credit limit that actually meant that one customer could have 12 14 different credit limits.

Gunther Smets:

Your aggregate that exposure could be much higher than what your level of comfort would be with that customer as well. So that’s where we were as well starting from. We have over a 50-year piece so we’re not talking about are we NCP company or we are not a local company. We’ve got a bit of everything. We’ve got five different versions of ACP that’s still out there around. We have 29 different JD Edwards instances or Oracle-based instances. We have around 60 different proxy systems that touch credit to cash somewhere during that process as well. So the challenge is how do you create that environment into one single source of truth. That’s really what the goal is or the ambition is but also the challenge at the same time. So our focus on the right-hand side of the slide while it’s the right-hand side for me and these is how do we go to an end to end process design. How do we stop silo thinking and a lot of those things as well? How do we focus on data? And I’ll come back to data because it’s your universal language.

It’s the only common through tech you have if you don’t have your data in the order you’re actually going to capture less than 50 percent of your value. If you start to board data garbage in garbage out it’s as simple as that. So how do we then also go towards standardized processes?

And I think the common miscomprehension is that the common or standardized process equals the same or is a synonym of a unique process which is not necessarily the same standardized process just means as you’re going to drive common capabilities across the process from start to went were a unique process means I’m gonna do replicate every time exactly the same thing over and over again. So if we standardize a process that means. We want to be customer-focused we want to have one credit limit per customer instead of a multitude of credit limits. So that’s a standardized process how we assess that credit risk that can be unique in a sense towards the type of business that we deal with because not every single business is different you go to market model will be different but it’s still the same kind of standardized approach that you would take at the back of it.

Gunther Smets:

So that’s a bit the difference between unique and standardized as well. And then it’s about how do we get to that technology framework which is really the emphasis of the presentation today on the technology side. I think we’re not like million peoples in this room so just interrupt me if you’ve got a question right. So you don’t need to wait until the end. So the first thing is the data challenge. And as I have said before data is the only universal language that you have within a company.

It’s every single data attribute that you would have within your company is something as an asset that you will use during your process as well. Anything that you want to do all the cool stuff that we’re hearing during this week as well. You cannot do any of it if your data is not in order. So that’s really where it starts. It’s the essence and the core of everything. So this is just highlighting what are some of those data challenges that we’re faced with because of being a decentralized company or where we started with.

So you’ve got all of those different process variations, every single process variation, therefore, has another definition of the data that you use but in the process itself. We have the same data objects being maintained in different elements. I think the simplest example of credit to cash would be for us a customer. We don’t necessarily have one golden customer account the same customer can exist in six different that’s a piece but different as a pin number allocated to where it exists within the twenty-nine different that naming can be a little bit different. You don’t have a unique reference so, therefore, every single data attribute to say how I want to aggregate up exposure at the customer level becomes very hard if you don’t have a golden account where you don’t have one master data set within there.

So that’s also part of the challenges that we have. We also have didn’t have really roles and responsibilities for how you set up your master data. We did not have one central group of master data. It was done sometimes within a commercial organization a customer service organization. Sometimes it was done in credit to cash. It could have been done all over the place on how the customer was being set up and the quality of your due diligence was also wildly different versus those businesses. So was it not really a nice place but we started from.

Gunther Smets:

This was quite a bit of a challenge that we had in there. So I’m not hiding anything. This is what it was. So our solution vision is if you look at it from a people process that technology I’m not necessarily going to go through all of these boxes here I’ll talk through it is from a people aspect we recognize and saying well first of all you need to have one strategy for master data. So we have a corporate data officer who’s actually there looking and saying how do we set it up from a people aspect.

How do we have one massive data group regardless of if you’re a salesperson and then a chairperson or whatever it is every single motion data component is being centralized from an organizational concept within one master data group? So where you then have the specific elements. I’ll come back to it as well and saying how does that fit within a technology design as well. Because if you’ve got all of these entry points that means that you need to have one consolidation or a massive data governance side where you’re going to have to store everything in and that everybody subscribes to.
So it’s more the aspect of the strategy that we’re taking and seeing centralize your data in how you create data in the first place from the customer a vendor a material master and then have the rest of your organization subscribe to that data but not create the data. So that way you have a much stricter and contained a governance model of how you start out with your master data, to begin with. So. Now the next slide is really busy from a technology aspect but that’s pure because of the landscape where we start from.

Right. So this is just an illustration of how our technology design or architecture is looking out right now. So this is what we’re building for a credit to cash aspect or is already in effect to a large extent as well. So the boxes that are labeled in TCC or D.C. That’s the different ACB instances that we have. So where we have a DC 2 which is for a food ingredients in our bio industrials we have a DCC the last C stands for commodity it’s for agricultural supply chain management that we have a specific as a b instance in there we have d that at the end of it that’s for our dissembling process. it’s for the protein business basically it’s for the kill site so the chickens and the cows or beef that we would have in there we have also our DC L which is coming on board and are PCL is more for financial markets. So we have more of a banking side towards the NAACP instance. So they all flow into FSCM. So FSCM stands for a financial supply chain management it’s rebranded by SAP right now it’s called receivables management but basically, it’s the same thing. And the only thing that that is that middle green box there that is just another sap box where you only switch on the receivables management element. So it’s another SAP box. And we flow in. So you’ve got back and forth to its point direct into FSCM compliant one hundred.

Gunther Smets:

And what we have done that all of the capabilities using also the HighRadius accelerators that sit within that FSCM and box to manage your credit management to manage your collections manage your disputes but also your cash application. It’s a one-stop-shop. The challenge. It worked at the beginning. It worked perfectly and it still works perfectly for an SAP environment so it’s your one-stop-shop regardless. But how many SAP instances that you might have where the challenge was is that you didn’t have a solution then for your twenty-eight twenty-nine different JD Edwards libraries that we had there.

So we did partner with a HighRadius as a proof of concept and that was. Last year I did the presentation as well and at that point in time, it was just a proof of concept. In the meantime it’s life. And so it’s working for all of our North America libraries already and we’ve got a face approach. So we’re taking the J.D. Edwards data or the Oracle data. We actually have an acuity layer which is converting all of your data from a J.D. Edwards format into SAP format goes into an oracle staging area and then there’s like an SAP data services it just pumps it into your single source of truth.

So basically it’s you create a one-stop-shop to have all of your data that flows in there. Now the black box that sits on the left-hand side of that is where you’ve got your master data governance. So that is where that all of your master data sits within that box and that’s where they are created. And everybody subscribes to that single box of the truth. So we do have API as well with external data providers like Dun and Bradstreet that help us with cleaning out the data. Doing also due diligence. So a lot of our customer base checking is automatically out of the MDG box or is out of the FSCM box.

So how we do our credit management as well is that actually coming into that green box so our score models, for example, I’ll come back to that. They’re automatically populated directly within the green box on the slide here. So we’re now on the road map to be 100 percent within that single box across our very diverse landscape within the next fiscal year. So which would be 2020. Everything’s gonna be in that single source of the truth which means that now in our shared service centers that they can just work in one system we have aggregated visibility on everything that we do and we can start mining the data the real value add work than from an analytical concept also comes with that. So that’s really the next step and how we can drive it. Yeah. I’m sorry. I could catch that and bring in me.

Audience:

The baby boxes that sort of your commercial data hub feeding in for your credit analytics purposes.

Gunther Smets:

Not really. So the BIBW Business Warehouse or the business intelligence business warehouse is our current stop that we have to do some of the analytics that comes out of it. We do have a direct interface with Salesforce.com as well with the CRM solution just to make sure that we’ve got that connection as well and the self-service component we’re working on a Web portal so that a commercial on the field can just ask for a credit limit and directly interfaces into here. We do also now have a data lake. It’s not on the slide here which is a Hadoop data lake where we dump everything from every single cell from a Salesforce force down from your order systems from your SAP element.

It all goes into the data lake and then we mined the data to power by or at Tableau or that’s how we would do it. Yeah. Thank you. So where for the moment all of the tools from HighRadius are still the accelerators that we use in here. So because we’re not in the cloud. This is not a cloud solution. However, we will continue now to upgrade to S4. So with S4 we will make the assessment as well and saying are we going to leverage some of the cloud solutions in there. Now the main component for us is of all of those elements is how do we leverage the data that we have internally. So and I’ll come back to it in a second here. Talk about some of the capability challenges and how we are solutions for them. So some of where we started out and I touched the way a few of them from a capability side both on credit and on collections the credit management was very judgmental. It was done in every single business they asked for financials they did some ratios that did some analytics they put in some quality developments but it was more art as a science and that element. So we don’t really have a common language on the risk if across the corporation as well. So how you would assess risk from one customer to another would be wildly different on how you would do it. We had very scattered risk mitigation options. So how you would mitigate the customer risk, on the one hand, we were under mitigate it. On one hand, we were undermedicated and on the other hand, we were overmedicated because we even found instances where the same customer was trade credit insured with three credit insurance companies exactly insuring the same risk that one business did not know of the other. So you don’t know what you don’t know if you don’t have the visibility. Then you can’t start improving it as well very highly manual ordered release process every single order almost in certain instances was touched manually to release it.

Gunther Smets:

So on average across the corporation, we had to touch to release the order that year around 40 50 percent of every single order we had to touch. So a very labor-intensive manual process not really a rule-based on how we were doing that element as well. And it was a very you read through an active approach. Everything was after the fact. Nothing was predictive. Nothing was bringing us value towards where we wanted to be as well. Now on the collection management side there it was collections you have a few things that you can do. So the simplest thing is your call for dollars meaning you do a telephone call you ask the customer when are you going to pay my invoice if you’re lucky you get like a payment promise and you move on that’s calling for dollars. But in my book, it’s not really collections in that element. We can have other people or other resources doing that kind of stuff as well. Plus what you see if it’s very just human and behavior-driven which without really a dashboard that you have people doing collections they’re gonna call the customer that was gonna pay you anyway. So it’s not really a smart collection strategy that you have in there. So you want to have a defined cockpit. Or work board that drives based on the rule-based that you have a smart collection scheme in there that people are really focusing on where to value add of those activities would be. Now. And you had different touchpoints by the customer as well. If a customer had overdue Smith or Coco business and with our starches business then potentially they could have two calls from two collectors asking for the overdue invoices which were going into the same accounts payable department creating frustration and saying I just had a call from your colleague. We just told you where we were. So these are some of the challenges on the capability specifically on credit and collections that we were faced with.

Audience:

[inaudible]

Gunther Smets:

Sorry. The question is who releases the orders in the organization. So it used to be a customer service organization that released it. But this is something that we stopped it immediately because it’s not really a commercial decision you want to have that segregation of duty between the two as well. So it was not a healthy place to be now but when we transferred all of those activities to the shared service centers as well. We have the credit team so we’re very much process-oriented and our credit team is split into. You’ve got credit analysts who do all of the credit assessments and the second part of that team is managing orders so they will release it. So they’re dedicated to it. They do nothing else every day to analyze the root cause analysis on why orders go on hold. Assess if they are allowed to release and work within the SLA. So if every individual company. So that’s how it’s set up right now. So capability goals First of all on credit risk management is to make sure that we are being proactive that we’re not running behind the facts but that we have some predictability that we can look towards the future and how we can do that.

For a lot of those elements. I look at credit collections or credit to cash which is a perspective that I really want to ingrain within the business within Cargill as well. We are not a cost center. What we are is a revenue generator or avoid revenue leakage within the corporation itself. It’s more of a mindset than it’s a and it’s anything else because if I’ve got the people working with that mindset they’re going to look at certain aspects also completely different and just calling for dollars or doing an assessment they’re actually looking and saying how can I enable that customer to further grow the credit management solution comes to a financing solution.

Gunther Smets:

So. That element is really looking and saying how can you then focus on a risk-adjusted return if you’ve got those customers that you want to push forward. Do you want to help grow? Then how do you make sure that you have your returns reflecting also that risk? Same thing if you would go to a bank you for a mortgage. The bank is going to make a credit assessment based on that. There’s gonna be a certain interest rate. So it’s how do you focus on risk-adjusted return on capital in the way that you start working with a lot of those things as well. And how do you have increased process efficiency? We have a lot of low hanging fruit. If you’re a decentralized thing you start centralizing and you leverage your scale. It’s there. It’s immediately there. Now three years down the road those efficiencies become a little bit harder to find as well. So you need to go towards technology and streamlining your processes your salary arbitration your consolidation of activities. That’s gone. So that value we captured. So now it’s to the next generation of value that we need to capture.

So on your collections we are looking at one thing is improving working capital but it’s not only improving working capital doesn’t necessarily mean decreasing your DSL just means making sure you get paid for it. That you do you have the visibility if you give 60-day terms that it’s reflected within the pricing methodology of your customers as well. So we partner a lot with the pricing managers within the different businesses to say how can we integrate the elements that we will get out of credit to cash within the pricing module that we would that they would have in the sales force SRM or wherever they store it as well. So it’s an end to end element as well. And the key aspect this we wanted to have an improved customer experience because we had a lot of feedback on their hands. Hey, Cargill is a very complex company to work with. We want to have one single point of contact not only for salespeople but also for the back office for accounts payable departments and so on. So how we can partner with that so challenges capabilities. So now we’ll go on to the solutions side. So how do we do with?

And I’ll start with the credit side. On it. So what we have is on that technology slides from a few slides ago we have that integration with credit agency data and we have an integration with an internal performance data and internal performance data is not only your payment behavior but it also how is the contractor performing the customer performing on the contract that they have with us. Are they picking up the orders forecasted versus what we anticipate that in or they slagging versus their forecasts which might be an indication that there’s something else going on as well? So how do you look at all of the data that’s available as well, so specifically for Cargill. We’re also looking at proactive for what is driving the predictability of risk. Now the interesting fact that might be different for your businesses but the interesting fact for cargo was that actually if you look we spend 80 90 percent of our time analyzing financials. Where the predictability of financials or a set of financials is less than 10 percent of what the likelihood of default is. So we’re spending 80 to 90 percent on something that’s actually not going to tell you anything at the back of it.

Gunther Smets:

So what we’re now looking at and saying what are the drivers for risk. So for cargo that is really looking at the environmental risk. So we’re an agricultural company. The crops are not on the field. The customer doesn’t have any money if they don’t have money they can’t pay you. So it’s logical that you’re going to see what is the environmental kind of risk for your customer as well. Another element is because we are a global trading company so we need to look at what are the supply and demand chains on a global basis. How are the soybeans stocks and reserves or how debt flowing? What’s the future position that we have in certain aspects as well. And how can we use that information within our credit scoring approach at the back of it? They’re not purely on financials as well. Another component would be in saying what is really the geopolitical risk. We trade in over 70 countries on a global basis of sovereign risk is a very big contributor. We’re going to treat Venezuela differently how we treat the US in that aspect which is a very extreme example but there is a lot of nuances across all of it as well.

I guess how this legislation blocks similar to those components as well. So it’s looking at what is what do you have from external data components so not buying a score necessarily from a credit agency whoever it would be but buying data elements and buying statistics and I led analytics so that you actually built a score modeling approach that’s going to then go into the accelerator that you have within the that we have within the FSM environment that’s going to look at all of those predictable elements in there. So we partner with Dun Bradstreet on that to not create the OMB score model but to create a Cargill score model to come to what uniform language of risk for Cargill which means that we can then compare one business to another business regardless that it’s in our grain business or in our cocoa business or wherever it would be on a global basis.

You’ve got one uniform language through. Now the other aspect there is. Which is the concept that’s on top? There is a normalized scoring. Now the typical way if how regular companies are looking at a scoring model if you implement a score model approach is that you’re going to look and say I’m gonna build up Sorry.

Audience:

Sorry, Just a quick question on Normalized scoring. You mentioned you partnered with DMB and rather than leverage their in-house scoring Scorecards the Cargill has a preference use their own and I’m just curious to understand why you would not take advantage of the expertise and in-house backtesting of their models that come with their package presumably instead of opting to maintain your own which then you have to do your own backtesting and all that sort of stuff.

Gunther Smets:

Yeah, it’s a good question and let me clarify because what we are doing with Dun and Bradstreet is, first of all, we’re buying their full data package so the data package is different individual data components that are in there. We do use their score model to back Test towards our score which we do with the rest of the industry as well. So we use direct analytics team as well internally within Dun and Bradstreet to say how do you go towards. So it’s instead of buying a product you’re creating a product. So it’s a cool code development with them. So instead of just taking your use case and then saying I want to have a scoring model I want to have this you’ll go under the hood and saying what I want is I want your random three I want to have your random forest methodology that’s underneath it and we’re going to do the regression together based on the data that we see in here. So you create a partnership from a learning environment it’s not necessarily that we do it on our own. Does that clarify? Okay.

Gunther Smets:

So the normalized element that’s in there that’s basically. So instead of building 60 different kind of score models for every single type of customer that we would have within the organization what you do is you build a data set that might be 100 200 data attributes whatever it is thing the raw score model and I got to the NBA expert here in the room is around 1300 data components that you’ve got in your score. So if you look at normalization means that you’re not going to penalize one score attribute that you can’t score. So if you don’t have any financials you’re not going to penalize the customer because they don’t need to give you financials you’re going to score at zero elements and still come to the same kind of output. The benefit of that is that you’re basically with one data set. You’ve got millions of permutations you don’t all of us and don’t have one score model. Then you’ll have a multitude of scoring models and that’s when machine learning then can help you at the front end of it and saying how do you make your environment learning at the back of it. So that basically you make a continuous circle of continuous regression of continuous backtesting based on your own performance data. Which is the clear benefit of what they see in an end state as well? If you do it with an external partner if you then backtest it if you would do it on your own landscape you can make it continuous. So you’re basically backtesting on a daily basis your own models at it. So you don’t run into something that would be too fixed or too static at the back of it. So on the collection side. In a nutshell, what we are trying to do here is really looking and saying collections on itself. The workbench that we have in there so the cockpit that we have in there should be fed by everything that you learn from a credit risk assessment basis or should be fed by everything that you learn from your cash application. Or from everything that you have from your disputes, collections are actually the repository of all of the elements that you learn from the root causes of the surrounding processes from an end to end perspective and should create a dynamic element in there.

Gunther Smets:

So if we talk about predictive analytics on a collections element we’re not actually trying to predict what the future payment behavior is going to be of a customer which is maybe a bit of a different approach and how we then look at predictive analytics the way that we look at what’s the future behavior of the customer is root cause-based. So what we are trying to predict and to focus on is seeing what created the issue in the first place and then focus on resolving that issue. So it doesn’t repeat itself. I’m not necessarily interested in knowing what that customer based on the current status what that customer’s payment behavior is going to be next month. But I’m interested in saying why did the customer then pay in the first place and how can I fix that. So with the next invoice, the customer is going to be on time. So how can you then leverage all of your data that sits underneath to have that predictive component? On the root causes not only a behavioral aspect. How can you fix the root causes and how do you make it end to end from that aspect as well. So from a collections aspect, we are looking at them more so the type of questions or the scripts that we would have for collectors are more focused on. Why was the payment delayed? What was the reason? Is it something that’s internal within your organization. Is there something that we can work on a process design as well. And it works both ways. We see that because we have a lot to improve as well so it creates more of a partnership as well. And I don’t want to pick on you Mike in the room but we have an example with the land the who’s also our customer who was then faced with some issues with getting payments from us. So if you look at it from an end to end perspective then you can start facilitating some of those components as well. So that’s really on how you start than creating that partnership power how that single focus of the customer then goes out as well. So we’re not looking at any of those components within a silo-based approach as well. I think. Those were the slides that I had. So any other questions.

Audience:

I got. Yeah. Right. So you know you’re using your usage you’re using it to really great unique way. One thing I was going to ask in the risk scoring models that you have for collections right to drive those strategies are you using external risk data to drive prioritization of data from Bradstreet to sort of help determine the collection strategies. Anyway.

Gunther Smets:

We do so which is the integrated model from because we use the OMB data as well then goes into the credit scoring models and the same data or the outcome of those core models is also feeding into your collection strategies at the back of it. So instead of just saying I did use the data only for credit management, you use the data for your end to end cycle in there. We don’t only use external data from credit agencies one very specific example is gargoyles got his own weather satellite because it’s where crop-based. So we need to understand what the weather patterns are on a global basis. So we’re feeding this in now and saying how can we then start predicting and saying based on the forecast of the crop yields that would be in there. How can that impact the ability to pay off your customers from credit but also from a collections aspect? Yeah. Thank you.

Audience:

You mentioned early on that you have quite a number of business units that all want something different. How do you get their Approval that all these different business units are going to be run through one shared service organization?

Gunther Smets:

A simple question is don’t ask for approval. The simple answer doesn’t ask for approval. Ask for forgiveness afterward. That’s the simple answer to it. The aspect of what I’ve seen that how best is if you show them why you want to do it what’s the capabilities. What’s the value. What’s the money on the table what’s the strategy what’s your North Star what does good look like if I would come in and say. My play is that I want to standardize it and I’m gonna save you a couple of million dollars in efficiency then they’re gonna say I’ve got bigger fish to fry you don’t understand my business. This is not a standardization play you’re gonna kill my customers and my business and so on. If I come in and saying well this is actually going to allow you to focus your commercial strategies as well towards the right markets that we can also do. It actually starts modeling where you want to focus on next. So if you want to be aggressive in a certain market thing and say. Basically, the end state would be is that we can predict what the default would be or what your profitability would be in a certain kind of area as well. So it’s more the sells the sales pitch like as a global process manager you’re equally a traveling salesman that needs to start explaining and saying what’s the value that you’re going to get out of it. And then either they buy or we force them to buy. And because that’s the other thing as well it’s not you buy or you don’t buy you buy because you believe in the strategy that we do or we need to convince you more. But it doesn’t change the outcome. So you need a strong mandate as well from a senior leadership team.

That’s going to help you. I think one of the biggest elements that I see overall not only in cargo but in a multitude of companies is the consensus culture. You need to have buy-in from any man and his dog before you can move forward. And it doesn’t really help you to accelerate a lot of those things. So the consensus is good if they believe in the strategy. Otherwise, he needs a good sense of mandate as well. Too much democracy is not a good thing in this context.

Audience:

So just a question around your comment on mindset. And sort of transforming the approach of your credit organization and presumably the organization of your collection as well. But I think you know from my standpoint there’s perhaps a better integration of the commercial aspect of. Credits roll in customer management. How did Cargill accomplish short of that change of mindset? You know the credit space has for a long time been around conducting risk assessments. What’s the risk. And now it’s with data. Some of the automation coming into play you don’t have so much of a need for that rule anymore and it’s more about alright I know what the risk is. What is the commercial implication of that risk and towards our strategic objectives? And that’s a meaningful change in mindset. Absolutely and so I’m curious to understand what I hear from you what Cargill did do to get your organization there.

Gunther Smets:

Yeah. What we did specifically to change that mindset as well because the first thing that you need to have is a seat at the table with a commercial organization which wasn’t historically always the case. It was a back-office transactional element looking at risk and not really looking at opportunity. So what we started. And especially if you move towards a shared service center because that is us versus them kind of mentality that that immediately goes on as well as we started to the stakeholder framework. So which means that we’ve got operational meetings depending on the frequency depends on what the criticality is of the business where we’ve got the credit analyst we’ve got the financial risk manager and we’ve got the commercial group into one room and saying OK this is the forecast of what we’re seeing in your portfolio.

This is where we see where your proportion of first gets or you start having some of those value activities as well. In the beginning, it didn’t always run that smoothly because you don’t have the right capabilities within your credit risk teams that can drive certain of those discussions or the commercial organization doesn’t open to it. But the format allowed us and also forced us to sit them at the same table and to have a discussion and really communicate that saying. There is value here. If you don’t get the value out that’s a people issue of the people at the table.

Gunther Smets:

And it has nothing to do with the processor with the data or that we’re offering you. So it’s coming and again having that also from that commercial sponsorship from an executive side as well and saying this is our value or part of our value play to devote the future as well. It’s equally educating your commercial organization and saying selling is not just selling your product. It’s looking holistically on how your customer is behaving and what the potential is of your customer as well. So we did a lot of coaching and a lot of training elements on soft skills with a lot of our credit people as well to see how do you started conversing some of those elements.

How do you have that type of communication with your commercial organization? So it’s an investment and it definitely is a journey before you get there. But if you stay on it and you have the format which for us was the stakeholder model, then we do see progress and quite rapidly and certain elements and once you get a few success stories and you published these success stories then the other groups are saying, “Why don’t we have that, why it is not working when it is working for them. Why is it not working for us?” So, that’s a bit of illusion that we see.