AI-Driven O2C Transformation: 76% Efficiency Gain in Average Days to Resolve Deductions

[0:00] Ganadeep Rey Patlolla:

Okay, so I mentioned that there was that recent acquisition that happened. And the thought process, because it was four years ago that we were a new company, a new Berkshire Hathaway company. It wasn’t a lift and shift where we took systems and people from PNG. PNG was centralized. So I came on board, we started from scratch, we hired new people, we had to establish processes and at the same time, we got new systems. And when I say new systems in a very short period of time, we put in a new ERP system, which is SAP. All that had a chair challenge, to begin with. But very quickly, we’re looking at opportunities like what we could do to make things better as we move forward. So one thing that I should mention is it’s kind of inside all that change that was happening when we started a new company. We thought we weren’t going to have enough people. And we did a little bit of outsourcing. That outsourcing took place for about one year. And then we took everything in. And when we took everything in, we had fewer people and we were trying to focus on what tools through automation systems, what can we do to make things better? So right here, we’re going to talk about the cash application process. When we started, it was a new system. We had SAP, we were applying cash and SAP, it was completely manual.

We had a bunch of people that would do this manually. We didn’t have auto cash, which is a term that a lot of people may know about. So what our cash applicators had to do is they had to pull down copies of the information from the banks. When I say information, these are the check copies and remittances that were given to the bank. They log into a bank portal, they would manually pull it down and be able to look at it on one screen and on the other screen, they would start applying the cash against the customers’ invoices. It was time-consuming because when you have a lot of customers, they had to aggregate all this information. So it wasn’t like you have one customer and they’re paying 100 invoices they had to pull in from numerous customers, get all the information together and start pointing the cash manually in SAP. The other piece that we’re talking about here is when you apply it manually, you have to learn when customers take deductions. We have customers that could take hundreds of deductions on one payment and they need to better understand why those deductions are happening and are able to go when somebody would manually have to code out. Is it a short shipment? Is there a pricing discrepancy? Or they’re damaged goods and so forth? Okay, so after all that cash is posted, it’s all coded what I talked about. Okay, so I think he probably heard that there are challenges that are very manual. Okay?

[3:29] Ganadeep Rey Patlolla:

You know, when you do things manually, it just leaves room for error. Everything takes longer and it’s time consuming when you need to resolve a customer deduction quickly too, because when you do it manually, and you’re trying to figure out what it’s all for, by the time you figure out what it’s all for. At the end of the day, you’re already behind the 8-ball and trying to get back to that customer and maybe collect on something that could possibly be collectible. You know, the other piece about this is, if you don’t apply your cash fast enough, your collection team has no idea what to collect because there could be money sitting out there, that’s there to pay the stuff that’s past due that they may normally reach out to the customer to get. And post automation benefits we were seeing were posting cash on the same day. That’s a no brainer. But before lunchtime, all cash would be posted in the system. It’s something that we kind of keep as our touchpoint. If something is not posted by 12, which means that the users found an exception, they had to kind of dig deeper. By 3 pm, it was posted sometimes, that was the starting point. When we went live now. It’s 9 am, everything successful gets posted in the system. By noon, exceptions are handled so that way it gives back the analysts time to do more value-added activities. The amount of time they were using, it’s now less than half the time of their day to kind of do cash posting and system hit rates. We wanted to get a really high number of 90 plus percentages of hit rate through automation at the header level, which means at the entire payment method level, down to the individual item level, we wanted to be plus 90%. And we’ve reached that point.

And, as I was mentioning, we’ve created multiple rule sets that handle customer based behavior patterns and also payment patterns. So that way we are doing optimized cash posting rules. Now that was not the end of the journey that I also wanted to take. So after the cash application, we were wondering what’s the next logical step in this life cycle? So in our experience here, we felt like the deductions module was your automatic fit for the second stage of your evolution of that debt reductions and cash posted. So until we went live with the deductions module, that real value realization did not come through. And the analysts, cash application and deductions analysts started to collaborate with each other stronger because now deductions would say, hey, cash application team, can you establish a rule that would do this so that my job becomes downstream easier, kind of you will see that in our following slides, but that I believe in this order to cash affirmation, add yourself first the biggest win for us was cash application and deductions hand in hand. This is the deduction process before we went live with HighRadius, I’ll let Steve speak to talk on that.

[7:18] Steven Wurst:

I was pretty comfortable in those chairs. That’s kind of a shame that I had to get up, but thank you. So customer deductions are just like what I was talking about in a cash application. Yeah, we had a system we were using SAP, but everything was automated. And we needed to find ways to do better because of the high volume, the number of transactions of deductions that were happening in the consumer products industry, it’s very typical that you know your customers will deduct a lot, especially these large mass merchant customers that we sell to. It’s not just a lot of transactions and time consuming, but there’s also a lot of dollars at stake. So the sooner we could figure out the customer may be wrong, the better the opportunity is to go back and try to collect that cash from the customer.

So, you know, this is kind of old for us. And when I say old, it was not that long ago, but it is archaic. You know, we had customers. They take a deduction, and then we have AR analysts, they basically would get all the information related to that deduction, and then they would start validating and figure out at the end of the day if that deduction is accurate? Was it valid? Was the customer entitled to that deduction? Or was the customer not entitled to that deduction? And when you have one customer that might take 500 or 600 line-item deductions on a check. As you probably could imagine, that’s a lot of work for somebody to do. Okay, so I could talk to you a little bit about the challenges. I was a little ahead of myself here because when I said it’s manual, you could see just about everything was manual. Okay? So it takes a lot of time. And there’s a lot of collaboration aggregation, a backup for our trade fund deductions in particular. We do reimburse certain customers for promotional types of activities. Customers sometimes don’t want to wait for that money that we would send them. So they take those deductions which could be big dollars as well. So post automation, Ganadeep pointed out about that.

[9:57] Ganadeep Rey Patlolla:

So post automation, one thing you have to remember is for deductions module from HighRadius to work, the heavy lifting, behind the scenes, the big winner is the CPA module, the claims, and POD automation. So, we had to ramp up claims and POD aggregation through multiple months. So like if I were to say the step was the same way and didn’t change anything but this step, the autoaggregation of the backup documentation. This step is where the POD agents would read emails, go through EDI, go through customer web portals to pull down information. Now, that was the non-value activity that the analysts were doing so much like hours a day. Now, once HighRadius did all the heavy lifting, the grunt work, put them all in one place, which meant that the analyst had to just work on the research. They would know based on the cash application rules, what would be potentially valid deductions, invalid deductions, and the unknown? Hey, look into this kind of a set of rules.

Now from there, instead of two branch points, we would have three branch points with model multiple write-off rules based on different codes, different dollar amounts. We wanted to save time and say- Is it even worth facing the small dollars, the pennies, the under $10, under $25 for sort of reason codes, and also if you had a valid deduction, we changed the process. We, instead of just working with the AR team, we brought on board the automatic management team to identify the valid deductions. We will establish a process that you can log into HighRadius’s system, see what are the valid deductions that you can start working to create the credit memos and the debit memos for pricing the variances, or SMD problems. And, again, the value work that the AR team and the data production team would be focused on would be working with the account executives and customers to kind of try and see how much of the invalidates can we recoup. I’m sure that is the biggest pain point from across industries across customers. Even if it is invalid, a customer wouldn’t always pay you. So you have to kind of work with them.

And we are trying to establish better processes and also work with HighRadius to kind of have other tools that we can automate customer collections on those items. It’s still work in progress in that way. What did we achieve after post automation? Within two months of going live again, there are different metrics that you wanted to track. It isn’t just one metric. People talk about DSO, talk about a number of deductions or the number of volumes or value of the deductions. What made sense for us to look at it in a different way? How fast are you working and resolving a deduction, because volume could change based on seasonality of the business, you could also have different kinds of dollar amount things. So instead of dollars and value, let’s stop focusing on speed. Let’s focus on where we can get the value back. So, just talk about resolution. Resolution means you have identified this as a write-off or it is a valid deduction. So that way, you are only left out with the invalids that you have to work with. So within two months of go-live without even completely coming to a stable steady state, we hit 69% improvement and after go-live and we have hit the steady-state, we’ve kind of plateaued out at 76%. So this is the same number put in a chart.

[14:16] Ganadeep Rey Patlolla:

If you see this one here, that’s an outlier. Don’t focus too much because it was a data dump. And this kind of throws us off. So just focusing on the month that we went live. We went live in April 2019, and April and May were our go-lives hyper care stabilization months, since then if you see the number of deductions itself doesn’t really change by a whole lot. It is the speed at which we are evolving. That’s about three times faster right now that we are resolving takes.

Steve, you want to take over?

[15:09] Steven Wurst:

Yeah. So one thing that I do want to mention, and maybe it’s obvious, but I think it’s important that we don’t lose this message. If you are resolving your customer deductions, roughly 70% faster, it’s a huge productivity opportunity for us as a company. You know, my people were not getting two things before. But now they had that time to do it. Okay, so you can look at many different ways, but it frees up people’s time. The other thing is, this calculates into dollars and cents. So if you’re turning your deductions faster, and if you’re smart, and you’re focusing on the big ones, in particular, resolving them that bet will talk into huge savings for you in deduction dollars. Okay, so collections. We had challenges before like I said if we couldn’t apply cash fast, and we’re doing a lot of manual work, we’re trying to figure out what you know customers paid with their deduction stuff if we lacked visibility of how effective we could be with our collection activity. So there were a lot of challenges along the way. We also do a lot of our activities these days. When I talk collections, it’s not about dialing for dollars, and some of you here may be experiencing where I’m going with this. Customers are asking us to go into their web portal. The portal is all the information that tells you when payments are going to be made, what invoices are going to be made and what happens – the collectors spend a lot of time going in trying to figure out that information. So A/R processes have a downstream impact on each other. Do you think automating the full suite would be beneficial for you?

So we had a look at what do we want to do? So actually, the first thought when, you know, again, I mentioned we’re kind of a new company. My first thought is, let’s automate cash. That’s a big opportunity. Then we started to look at what else we want to do? In May 2018, we signed a contract that we’re going to deal with more than the cash application and see how it goes. So, in December 2018, we went live with the cash application, what we’re talking about, and we also have claims and POD modules that we’re working on as well. In February, claims and POD modules went live, the collection module went live, the collectors are able to go in and start using automation around that. Then, in April, the deductions module was working, and we stabilized all this in June. At this point, we have a cash application, we have a collections module, we talked about deductions. We have 65 AI agents doing work for us. So we’re pretty automated. We’re adding 15 more. I think you could hear that a lot of this will reduce a lot of manual work. We had success. We started in North America. We had success with this project. We’re seeing a good return on our investment. And now we’re rolling this out to the rest of the world throughout Duracell. So we are on the part. Are there any questions?

[19:26] Questioner:

Yeah. Hi, I’m Nick Flow. I’m with the GPO team with Apple. Thank you today for the presentation. It was great to hear about your journey and what you’ve been able to do. When you think about these couple slides you had on just before and the way you rolled it out. You went out with the cash apps module first, and then you did CPA. Why was there a gap between you when you rolled out CPA and the deductions module?

[19:54] Steven Wurst:

Ganadeep will elaborate on it. But, first of all, we went with the advice of HighRadius. So we made the decision that we’re going to get the suite that I just talked about. But I think when we thought we were going to get a lot of opportunities around the cash application. So let’s get going on that. And that’s kind of the very front the process when you’re going through order to cash and around the ERPs. I don’t know, from a technical standpoint, is there any other reason?

[20:26] Ganadeep Rey Patlolla:

So the production module needed more development work from the SAP, in the ERP space also. So cash application, SAP has out of the box, some solution that HighRadius uses. But for the deduction module, we had to build something. So we needed more time to build that. Now, in hindsight, we could have compressed the timeline somehow, but at the same time, it was also a burden on the project team from the business side. It is the same group of core people working on collections, cash applications, and deduction management. So we had to kind of think about resource burdening and also IT speed.

[21:16] Questioner:

And then you used phase one and phase two for CPA. What were the differences between those phases?

[21:21] Ganadeep Rey Patlolla:

In CPA, there were some challenges. I’ll be kind of honest about it. If you go and read any customer websites, terms and conditions, they don’t necessarily allow RPA agents. It’s a gray area, some of them don’t care, some of them care. So we had to go through our internal audit purposes to say everybody is moving towards RPA. The terms and conditions, the legality, they probably have not caught up, I won’t name the customer but then there is a big box customer who does not allow RPA agents for us to log into, but they also use our RPA to do other things. So it’s something that we were very concerned about as a legal process problem. So we went with EDI and web automation first and worked through the compliance and audit purposes for RPA for websites, and then we took it as a wave too. And also, it takes up a lot of time for the analysts to spend time with HighRadius consultants to go through step by step navigation of an RPA agent automation. So we can spend many hours doing multiple modules at the same time.

[22:43] Questioner:

And then the last question, I won’t take up all your time. So you did America first and then you did the rest of the world. If you could do it over again, would you do a big bang approach? Or would you still do the phases as you did on the slides?

[22:57] Ganadeep Rey Patlolla:

I can answer that, Steve.

[22:59] Steven Wurst:

Could I just get one more piece? And before you answer that, I think one more piece. That’s important to know about your first question. And you talk about that timeline. One challenge is that I have two great people that were on the project. And it’s not like I quit, I had other resources. I’m going to put these people on the cash app simultaneously. We’re going to do deductions and so forth. So we had to go linear because of resources as well. I wanted to make sure this is important to us. We have the best people working on it. And we just thought pragmatically, this was also the right approach based on resources. So Ganadeep.

[23:42] Ganadeep Rey Patlolla:

Yep. For the global rollout, that’s definitely I would recommend a phased approach. At least take one region, your heavy hitter as the pilot face. They had different processes but they were waiting to see how it works out in HighRadius. So, I’m sure it’s in every corporation but I think also, each region has different payment patterns and habits and customer patterns, I think North America because of the lockbox methodology, and also how the banks are mature, like collaborating with JPMC, HSBC, BOA, all of them they are very mature in working with HighRadius. So it kind of felt natural for us to just go over the USA first and when we are right now doing the designing and the blueprinting for us, the world, Asia has different complexities. If we had taken up that we would have probably not set a baseline model. So what right now is that we have established a solid baseline model and from there we are saying- Okay if this is my baseline, what is my departure? What is my delta for regional specific cases?

[25:12] Questioner:

Hey, my name is Murtuja. I am in Sanofi’s global team. When we talk about deductions, I assume you are also considering scoped disputes. And it is not only financial disputes but it may also include some other disputes like logistics or pricing that kind of element. So, could you please share your experience? How did you partner with other functions like customer service trade, etc? And what is your scope in these solutions?

[25:55] Steven Wurst:

I’ll just take one quick shot and then Ganadeep, feel free to chime in. So when you talk about disputes just so we know, it’s all kind of synonymous, each business here, each person may use different terminology. So when we talk about the customer deductions, sometimes it’s disputed, sometimes it’s short payments. But when these deductions happen, there are different owners in our process. So the accounts receivable team and I’m hoping I’m answering your question correctly in terms of ownership in the process. The accounts receivable team is actually responsible for gathering all the information for the deduction. Okay? And they need to identify what it’s for. Again, what you just saw in the slides, we automated a good part of that. Sometimes there are some kick-outs but the ownership falls on accounts receivable.

Now there are pieces of the process where customer service needs to get involved. So customer service, for example, they have ownership around certain deduction types that are unique to customer service and they’re more knowledgeable about. And I would say, in particular, if we can’t get the proof of deliveries, if it’s shipping related type of problems, we have them engaged. And we do have kind of a workflow, you could probably say it is because, in HighRadius, it allows you to assign ownership to the people that need to do the work on it. And we use that and we find that very resourceful. And whether it’s customer service or the A/R department, if we find that the customer is correct, and we actually have to issue the credit memos, the process is much quicker and more effective than what we’re doing in SAP. So customer service is using HighRadius for certain pieces that I described. And their feedback has been very positive that this is saving them a lot of time. So, you know, it’s an A/R. It’s an A/R software, but you got to think in terms of the whole stream, the whole order to cash stream. It does touch other areas and it offers productivity and other areas going deep. Is there anything else you may want to add?

[28:18] Ganadeep Rey Patlolla:

So for the disputes, for logistical troubles or even policy things, we rely heavily on customer portal information, which we kind of automate through CPA like claims and POD polls. We even work with our logistics partners to get signed POD to kind of backup our information and end up backing up the stream where we are starting this off is based on customer remittance codings. Either on their website or through the lockbox remittance payments, we are trying to identify every single coding value or prefix values or suffix values to kind of say, what’s the pattern? We want to identify patterns upfront to say that this is a potential or this is a strong candidate for logistical charges or damage charges and such. And we use the CPA agents to do the grunt work. I hope I answered the question, but if not, we can definitely discuss it further and more. But please ask us more if you have any questions.

[29:30] Questioner:

No, it sounds good. So that means apart from different interfaces or robots or whatever you are using to capture this data. So that means what you can anticipate as a deduction in terms of the dispute. Sometimes customers can call and tell you that there is a problem. So customer service generally locks that as a dispute. So that means customer service is using two different systems to log that information in your case?

[30:11] Ganadeep Rey Patlolla:

Not really, customer service for all dispute management, HighRadius is a system of management for disputes. We don’t use FACMS, SAPs for dispute management anymore. We disabled that was the past state. So the order management team logs into HighRadius to see potentially all the customers if they made a short payment. They can see everything over there. So the research as we are trying to say is a collaboration between the AR team and the order management team. The order management team relies on the various reporting that we have in HighRadius to make their work faster. So they just download reports that they can get, then they can start researching from there. And all the backups from the portals are also there so that they don’t have to, again, log into the portal again, or see the emails or the EDI to find out if there has been something from the customer on the same information.

[31:17] Steven Wurst:

Yeah, you know, we’re pretty automated, I’m sorry to interrupt, but we’re pretty automated. We’re pulling in all the customer deduction, the backup into HighRadius. We got the invoices with what would be like a customer ledger and we have proof of deliveries with the carriers, they get uploaded right away. So there’s a lot of information there that’s making the job easier for customer service. And I think I mentioned that the system has flexibility when I said you could kind of assign it to somebody. I guess it’s a little bit around a business decision to decide who that might go to? So is it your average customer service person or order management person? Where do you want to assign this responsibility to just maybe to people in your company and this is their full-time job. We were having this discussion yesterday, kind of a philosophical discussion. Systems sometimes don’t offer that flexibility and you got to build your business process around the system. This has some flexibility where we can make sure that we assign the work to be assigned to the person that really needs to work it and it’s making their job a lot easier.

[32:45] Steven Wurst:

Thank you.

[32:47] Ganadeep Rey Patlolla:

Thank you.