Automation Guide for Treasurers: Challenges and Expected Benefits with AI: Case Study

Speakers

Paul Watters

Director, Worldwide Credit and Treasury,
Mercury Marine

Amber Thompson

Treasury Consultant,
HighRadius

Transcript

[0:00] Anchor:

Thank you for joining today’s session on challenges and expected benefits with AI and MercuryMarine. We have with us, Paul Watters and Amber Thompson. Paul is the director of Worldwide Credit and Treasury of MercuryMarine. Paul moved to the US in 2015 to resume his current role as director of Worldwide Credit and Treasury for MercuryMarine. And we have Amber Thompson with us, Treasury consultant and HighRadius, Paul and Amber, the stage is all yours. Yeah.

[0:44] Anchor:

We also have a signboard with us. She’s going to circulate a signboard. Please write down your name and email addresses. It will help us to give you the presentation slides on email and also help you gain one credit for CTP recertification. So yeah, please don’t forget that.

[1:07] Paul Watters:

Good morning, everyone. So, my name is Paul Watters Hi, I’m employed by MercuryMarine, I’ve worked in their business for about 15 years, doing a number of things and opportunity kind of move in 2015 for the US and take up my existing role. So today what I want to talk to you about is an automation guide. I’m going to be talking a little bit about cash flow statements and a little bit about that actual operational cash flows. So kind of off today. Just a brief overview of what it is that we do, give you some context for the presentation today. The objectives for cash forecasting, what is it we’re trying to achieve with an accurate cash flow forecast? We want to look at what our existing processes and on how suspect there’s a lot of you in the room here today if you’re involved in cash flow forecasts, people are still using spreadsheets and summaries from prior years and so on written on the best practices on paper. I know that certainly what we’re doing at the moment and hence why we’re looking for a better way to do that.

[2:21] Paul Watters:

And then we’ll pass over to Amber who’s going to talk a little bit about what the hi radius solution is, and you know, what we’re really trying to do, what Mercury is trying to do to get a more robust and let some kick you cash forecasting outcome. So about our business, we’re leading manufacturers and distributors of marine propulsion systems, which is just engines for boats. So we do inboard and soon drive engines as well as outboard engines for boats. That is all we do. As of April last year, we used to have a fitness play as well with Life Fitness, but we found that off so now we are just additional manufacturing business but for engine parts and accessories, engines are about half our business. The other half are parts and accessories. So. So that’s what we do. We have distribution facilities across North America, also throughout Latin America, Asia, Europe. And so we also do have distributors throughout the Middle East to and so we’re, we’re a truly global private in terms of manufacturing, we’re really China and us centric, most of our engines are still manufactured in Wisconsin. In the US. We have a large customer base, we have 34,000 customers, which obviously creates complexity, both with incoming cash flows, but also with thousands of suppliers as well. It creates credit challenges in our payables and disperses material as well. And global revenues exceeding $3 billion. So we are one of the oldest companies been on the New York Stock Exchange for over 150 years of parent company Brunswickers.

And, so it’s fair to say we’re quite a conservative company too. So just having a look at the Treasury landscape and what it is we do today, so you’ll see up there some names that probably don’t mean a lot to you. They are there are businesses so MercuryMarine is our propulsion business and the largest of the businesses, land, and sea have about 18,000 customers in that business across North America also have added water as well, which we like to refer to as a product business. So that would make things like the lights and so on that go into boats, so our boatbuilder customers will purchase that product from us. And we’ll use that to build the boats. And recently in August of 2018, we acquired an electrical components business called power products, and that’s really driving us to get more automation into boats. So now electrification of boats, being able to control boats through mobile phones and all the things that people want to do today. So in terms of What we do here across those businesses, we roll up cash forecast for those businesses. And we, we consolidate those and send them up to our corporate parent friends with Corporation. And so you know, of course, across each of those businesses, we have cash inflows and outflows. And really what we’re going to be talking about today is, is you know, what we might look at on the offer on the cash flow statement and say our cash flow from operations. The working capital changes in payables and receivables will occur over time, we’re going to largely ignore the investing and financing pieces. But they are relevant because cash forecasting helps us to fund those decisions around capital expenditures, jewelry purchases, and also other activities that don’t necessarily relate to the day to day operations of the business.

[5:53] Paul Watters:

So when I look first at the objectives of cash forecasting and their short term objectives, somewhat aligned with their long term objectives, but they are different in some respects, we want to minimize our idle cash. So we don’t want to be tapping out facilities unnecessarily when there’s money coming in, that we didn’t know about, because we hadn’t forecast accurately enough. We also want to optimize through those through that data, obviously, our short term borrowing and lending decisions. Yeah, if we borrow tap one of those facilities on short notice, it is considerably more expensive for us to use those facilities than if we have even a week or two’s notice. So the importance of cash and the advice being accurate has a genuine impact on the business in terms of costs. Over the long term, of course, it’s not our profit that pays the bills. It’s not profit that assists in investing in new research and development. And so we need that cash if we’re building new products and so on. We’re quite a capital, intensive business. And so from time to time, we released a new product recently. And that R&D project over four years cost us about 170 million dollars. Now, we need to generate cash to do that. Our cash flows and accurate forecasting for others assist with other decisions. Yeah, like I mentioned capital spending, share buybacks, dividends for their investors demand from it, and so on. So it’s important that we get those calculations, right.

[7:33] Paul Watters:

So have a look at our current forecasting across categories, we’re going to focus on the AR but our process at the moment is subjective. It’s reviewing from a receivables point of view. Okay, what did I sell? Who did I sell to? How much did I sell last year? And how can I use that to guide me as to how much cash gonna collect this year? For our payables, of course, it’s just the reverse. Who am I buying from? What are the terms i? Is it large capital projects that I’m doing? Or is it day to day spending on stationery and payroll and those types of things? So with that, I guess the point of this is that for those, we’re doing this largely manually. So it’s spreadsheets. And if I put the data in front of a group of you, you don’t give me a different answer. And you don’t have reasonable justifications as to why that was. So what we’re trying to do working with HighRadius is to find a solution that’s data-driven, and that is not dependent on how somebody feels that morning. So, obviously, there are challenges with our current process, it is somewhat inaccurate.

[8:43] Paul Watters:

You know, as I said, we depend on who we’re selling to, it also depends on what time of the month we might be selling them, whether that cash is going to come in this month or next month, and it’s also inefficient as well. You know, we have people running around trying to work out our payables department. How much is coming through this week? We have any capital projects were spending may have been approved and invoices might have been approved that will move the needle. Are there any refunds that are going after customers this week that might impact their payables? And we’re doing the same thing within our receivables area as well, largely using as a site past, past trends and subjective assessments come up with a forecast.

So as we look across the various pieces on the last couple of slides, if we look at capital expenditures and acquisitions, we can be somewhat certain about what it’s going to cost us, certainly our dividend payments, we can be reasonably to note as well, and also equity being issued or repurchase. That kind of activity where we have a reasonable level of certainty over them we can plan for. With that, I think, these two, the AR and the AP are the toughest categories, I think to assess and to predict among those different categories that we just looked at. But their importance, it can’t be understated because they finance those other decisions that we’re making on the capital spin. And if we don’t get those forecasts right, then it’s difficult for us to invest in our business. And there is a risk, that the cash won’t be there at the time that it’s needed.

So looking at some of the factors that impact forecasting accuracy, and that also makes it challenging for somebody who’s sitting there with a whole heap of data, trying to work out what’s going on in this space and when the cash will be coming in. So on the AR side, we have sales variability, I could say that if we’re going to sell 100 million dollars of product this month, then we sell 110. Well, obviously that means I should be expecting more cash to come in the door. Conversely, we might sell 80 million that month as well. So sales are a critical input in terms of how much we’re going to collect in a given period. Who am I selling to?

We have, I’m embarrassed to say across our business hundreds of different things. Some of them are actually quite similar. It might be, 30 days 1%, net 45. And another one might be 29 days. Yeah. 1% net 40. So, who I’m selling to, and my mix of customers and mix within the channel as well will determine how quickly that cash is going to come in. What am I selling to them? If I sell an engine to most of my customers, it’s on the 15-day terms for the invoice. If I’m selling parts and accessories, that’s 15 products. So that mix of what I’m selling impacts my cash flows as well. Will they pay and take this, can’t really, so I don’t know. And so I sell two things, how much money they’ve got, depends on what their current strategy is with their cash flows as well, or will they pay me? So I can have some level of predictability about who’s going to pay me life because I think most of the time, it’s the same people but, you know, as I look across 34,000 customers to try and determine that I’d be fooling myself because I couldn’t get any great degree of accuracy on that.

[12:04] Paul Watters:

Another issue for us as well as rebates, we issued many millions of dollars worth of rebates in August. So from that, I need to determine whether my customers’ accounts are going to be putting credit or not. Because if we’re going to be putting credit or it substantially reduces their balance, they’re not going to send the money when they don’t know anything. So, for our payables department, it’s the reverse side on rebates, they’re going to be trying to determine if you issue $25 million in rebates, how much are we likely to pay out within the first week? And typically, we know that for a ground business in August that the net 30% of that so we need to be keeping in mind all of these factors. When we think about it from the payable side. What’s my purchase volume? We need to be thinking about who am I buying from because of course, we have different terms with our suppliers as well. What am I buying, is it a capital project? When will the goods be delivered? You know if I’m trying to connect inflammation from our procurement department to determine when invoices will be coming in, I need to be able to determine when the goods are going to be delivered. Maybe there’s a large order out there that we’re waiting on, that I’m planning for, but the person in procurement knows won’t be here for three months. That affects my cash outflows honestly because I don’t have an invoice. I went behind, will I pay in full? Or will I take this year? You know, what’s your strategy they have a strategy of conserving the cash? Or is it a strategy to maximize the discounts and earnings and sacrifice the cash and so it’s important to have a strategy around that as well. When will it be approved for the invoice as I mentioned that I offer a discount and will I take the discount? So a whole lot of variables, most of them on the AR and AP side of course and just a reverse of the other side.

[13:58] Paul Watters:

So our existing AR forecasting process, we do it annually. Each year in October as we sit budgets, we also review it monthly, and then we review it within the month as well and refine those forecasts week by week. So on the AR side, I’m looking for our cash flows and theme week today. What does that like the domain for the rest of the week? Am I expecting a large payment from a customer enough? Again, in their weekly starting on the right, a monthly forecast services bleed into weekly based on sales inputs. So it matters to me, not just how much I’m going to sell within the month, but when I’m going to sell it. Because if large orders go out, as they often do at the end of the month is where salespeople are chasing the targets.

[14:40] Paul Watters:

I need to be cognizant of that because that money is not going to come in this month. It’s going to come in next month. And then also, as I mentioned, having a look at the actual cash flows and sales updates, the sales area and do we think we’re on track for our targets. Is there anything large that’s coming up in the next couple of weeks? But all of this is subjective. And that’s what we’re trying to get away from. Monthly forecast, we do that, obviously, it wouldn’t be called the monthly forecast. And we do at 12 weeks in advance. So that’s what we’re looking forward to being able to guide ourselves in 12 weeks. Obviously, the further we go out with any system, the less accurate it’s going to become. So we’re really looking to maintain accuracy within a monthly period of beyond because there are just too many moving pieces.

Sales from last year is taken into account and will adjust all the way depending on increases or decreases in sales. So the reasons for the unpredictability within our business and seasonal trends are obviously affected is not too much voting going on in New York at the moment, for example. And so in Florida, certainly there is but we’ve got to think about when we’re going to be selling that product to people when our builder partners will need our product to build boats, and so on. There’s also customer payment variability as I mentioned, obviously sales and then rebates as well, waste factors among many others certainly influence the forecast. So as we look to some more of the challenges, we have the data there. But you know, spreadsheets aren’t really a very accurate way of us measuring that. Still very subjective. You know, as we’re watching data more so than ever now, the question is not so much getting the data, that how do we use the data to drive business decisions. And quite frankly, we don’t have expertise or tools. In that area, we don’t have employment data scientists to enable us to analyze that information.

[16:40] Paul Watters:

So again, that’s part of the reason we’re looking to do things a little better. Looking on the AP side, not dissimilar from the AR side, we’re looking at the purchase as last year, we’re selling more engines. Obviously, the input in terms of procurement and so on are going to be great at the most. Conversely, if we’re not building as many engines, we can expect the volume of their purchases, and accordingly our payments to be down. So we do weekly and monthly forecasts within that AP process as well. And it’s largely done based on data from the update. But if somebody comes to our payables department and says, “Hey, this capital project, I’ve just been holding them off for two months, we’ve signed approval today. And I promise to join that I’m going to get the money tomorrow.” Well, that’s not something we can forecast for. And so it’s important we have good processes and expectations around what is an acceptable standard for that kind of activity. Reasons from predictability on the AP side, again, we’re a seasonal business that we have vendors that are categorized into homogenous payment groups, we have probably as many times been from the payable side as what we do on the receivable side. They vary by department groups, and I have pictures for each of those words. So when we’re forecasting AP. As I mentioned earlier, and your rebate program, we’ve got to try and determine how many of these customers that were paying rebates to will end up requesting from us a check payment. And there’s a number of other complications in that as well. I’ll just work that credit off. That’ll be fine.

So, there’s certainly a lot of complexity in payables as well. I think one of the things about payables too, is that at least in our receivables, at the start of the month, we have a portfolio that we look at, and we can be reasonably assured that most of our cash is going to drive from the portfolio as we look at it on that day. In payables, we have all kinds of terms that are coming in, and we have invoices of course that are coming in throughout the month as well. So again, it matters who we’re buying from, and what our terms for those customers are. It also matters about commodity products. Being an engine manufacturer, we buy a lot of oil, and that oil is typically paid on very short terms. So I can’t look at my monthly cash forecasts and determine with any great extent what those oil purchases and payments will be because they come in and they paid within two or three days. Likewise for our transport, commodity products, low margin typically paid within the month as well. So they are inputting challenges as well and of course, some of that or a lot of that is determined by what we’re selling, where we’re selling it to know what the trade arrangements are for those different things.

[19:36] Audience:

For these short payment terms, it is difficult to forecast and so how do you conquer that challenge?

[19:43] Paul Watters:

You really need to be working with procurement things closely. This is the reverse equation for the receivables department working with sales. We need guidance from procurement who are making these decisions about large items that we don’t have any visibility to until we receive the invoice.

[20:06] Audience:

Is that input manual or is it some something you would put in a system?

[20:11] Paul Watters:

That includes the manual. And so I think the easier part of this is to determine repeating exposures and repeating payables items, payroll, things like stationery, printer consumables, those things that we’re typically buying every month and every year. Where the challenge comes in, I think, is if we have large capital projects that might bring out a lot of payables in a given month, we still need to be working with our procurement and project teams to ensure that we have an accurate input because without that input, we’re really, really busy.

So looking at how forecast accuracy impacts our overall business. Debt facility drawdowns on short notice I mentioned. They increase their borrowing costs. Now we want to be operating as efficiently as we can be, and to do we need accurate forecasts. Also, though, and perhaps more importantly, if disrupts their internal planning, if the money is not coming in or going out as we anticipate it will, over time. We believe in public companies. If we’re guiding markets, the one outcome, and we’re producing something is significantly different. That’s obviously something investors don’t like. And so we’re looking to guide and keep those promises around for cash generation, we will form a career in the same way we’re looking to guide earnings and to meet those targets.

So operational efficiencies in the current process, as I mentioned, we use Excel spreadsheets and I’m sure we’re not the only ones there. Collaborating with multiple teams obviously together, that information is difficult. It’s also subjective within the people who are giving me forecasts as well because not only do I have forecast from within the US, but I have other international businesses as well that will provide me with their forecasting. Actually going back to it. And so all of that put together basically constitutes our cash forecast. So doing that in Excel is not easy, obviously, or it’s inefficient more than nothing. But manually consolidating these multiple inputs as well as subjective inputs from various people around the world.

[22:23] Paul Watters:

So, yeah, again, we want to be guarded with these decisions by data, not by the way we’re feeling in the morning or by somebody’s subjective decisions because I’m sure everybody has a good reason for what they’ve come up with. The question is, is it backed by data? And can we use history at the moment to Excel spreadsheets? So, operational efficiency inefficiencies in terms of the existing process. Obviously, this information goes to our board, and it also goes to CEOs, CFOs and other stakeholders who have an interest in it. So sharing that information Via via spreadsheets and so on is antiquated. So it’s obviously hard to clarify back and forth interactions with things as well.

We’d like to have the one platform where stakeholders can go into the various places, and they can have a look, what’s the cash forecast look like this week? And so that’s part of our idea as well. So can I get a show of hands for those who are involved in cash flow, who are actually using spreadsheets? Okay. So hence, this is probably a good reason for the product, right? I think it’s certainly been something we’ve been talking about for a long, long time, and you’ll hear it regularly come up within forums and so on with people asking about what methodology is that you use? And I think at the end of the day, most of them are using largely subjective methods, methodologies, driven by spreadsheets, so there’s a limitation. Difficult to collaborate across teams, as I mentioned, simple formulas, I can’t look at a suite of three years history with millions of invoices both in payables and receivables, and come up with anything that is particularly compelling in terms of a data-driven process. So, that’s where I guess data science and also, the systems that we’re talking about today come into it. So it’s also manual. It’s time-consuming to waste time. There are rekeying data. And having that data shared around for somebody else to repay is the ultimate in low value worth in my opinion. So we should be looking to do things more efficiently than we do today. So with that, I will hand over to Amber.

[24:45] Amber Thompson:

Thank you. So I’ll go over a little bit about our HighRadius solution. First, a general overview on how we are implementing for MercuryMarine. We will be pulling in the end state of their operational costs of AR and AP. Yeah, they’re using Oracle for their ERP. So we are pulling in accounts payable and accounts receivable into our cash flow cloud, our forecast cloud, to be able to automate the process and allow everyone to be able to access the system, whether they’re viewing this forecast or adding information to it. And we are leveraging AI to be able to automate and make their forecasts more accurate. And again, we went over it a couple of times about why we’re going through AR and AP. And that’s because these cash flow categories are the most variable with MercuryMarine. So now I’m actually going to show you a demo of a demo system that we have that will be providing to MercuryMarine soon. So this is the page that you get to when you aren’t trying to look at your forecast over. No, it’s not showing. Sorry, guys. Okay, can we see now? Okay.

So we’re going to try to turn the lights down just a little bit, it’s a little bit easier to see. But this is your forecast overview in HighRadius. Our first card of talk here is the forecast currently showing for the next three months, but we can have this period go out as far as six months. And we can also alternate between this weekly bucket or a monthly bucket. We have a couple of other key features that we want to point out with MercuryMarine today and one of those is that we can analyze the accuracy that you have in the past that you know what kind of confidence you can have in your forecasts in the future. So to do that, we have these cards below, we have a forecast versus the actual in the past period that we’ve selected. And we also have a variance which shows the difference between the forecast and actual as both monetary value and a percentage of your actual.

[27:25] Amber Thompson:

So let me walk you through an example of the latest date here. On January 20, this forecast would be 67 million, whereas the actual is 57.9. And when you come over to the variants, you can see that that monetary value is 9.2 million dollars different and variance of 13.7%. This gives you a really good quick snapshot overview of what your variance looks like when you’re looking at a forecast horizon of one. So this is when you’re looking at your forecasts in just one week ahead. If you want to look at the values a little deeper and go through the numbers themselves on these variances, you can actually click open one of these cards and be able to see all of the previous forecasts that we’ve made on these weekly dates, as well as the actual cash flow that happens that we loaded into the system previously. So you can see where these numbers are coming from these charts, and so forth.

[28:35] Amber Thompson:

Yes, this percentage would be forecast accuracy, comparing this forecast to the actual. So on December 30, we forecasted this $50 million value when it was we had an actual for the date of January 20 to 26 of 57.9. So we have a resulting accuracy of 14.15%. Another key feature that we wanted to bring up was that we’re not taking the human element out of our forecasting. And we start with the AI-powered forecasts, but we allow users to also come in here, open up this sheet system and be able to add their own information that may not be something for any items that may not be forecastable from previous historical data. So if you have a customer that’s providing a payment, that’s a little extreme, if Walmart per se would give you or call you up and let you know that they’re going to be paying you a $50 million amount on Sunday, you can come in here, make that amount. Sorry, type in that amount into your forecasts, and you should be able to come back to your forecast overview and override, include this manual override to be able to see that this value changes here. It’s as simple as that. Do you guys have any questions while recovering the demo screen here? We’re going to try to bring the microphone.

[30:24] Audience:

If you did do a manual override, do you have the option to see the comparison with what was forecasted? Like the variance between was your manual override more beneficial than what was forecasted by HighRadius?

[30:41] Amber Thompson:

I think this is something that’s currently on our roadmap, but I’m not sure.

[30:49] Speaker:

That’s something that becomes part of the analytics area where you can get different reports generated but have a quick view here. You can just play around with the graph and see the difference of forecast versus actual, you can see there as well with and without many lower rates, that will give you an understanding of how it works.

[31:16] Amber Thompson:

Good question.

[31:21] Audience:

Are you allowed to or is there anywhere where you can document what the manual override was? And does the system hold that?

[31:29] Amber Thompson:

So this is actually something we are still currently building on our system a lot. And this is something that we are hearing recently from our clients and prospective clients as an option. So my understanding is that this will be something that we will be planning on putting on our roadmap.

[31:50] Speaker:

So the system does two things one definitely keeps a history of it. Okay, that’s definitely there. And apart from that, within the high streets in also be into input comments as well. Like, for example, we’re putting a manual override, you want also want to remember where the hell did you put this? You know, you can record some comments stating that “Hey, I got this fall or a promise to be with the change of plan or something.” But also gives you an option for the comments as well. So that at a later point of time when you’re trying to do a drill to of why these numbers will change, where there was a different budget forecast versus the manual overridden value, you can get a history of all of that. Yeah.

[32:43] Audience:

In your example, you had 50 million come in from Walmart. How would you know where that 50 million was forecasted to come in? Is there something that’s automated where if you plug in the 50, you can have that removed from when it was originally forecasted to become?

[33:02] Speaker:

So, this is something that we are unable to show it now, but within this forecast will have an option to do a drill-down. Right? Essentially, like for example, if the system shows that there’s a 10 million or 30 million that’s coming in the next week, you also want to understand what are the invoices that are part of this forecasted value. The system will allow you to drill down into the invoice level so that you can see which of these invoices are expected to be paid in this way. And then you’ll be able to manipulate it right there, where instead of doing this manual input, it will also be able to say okay, change the expected date, so you are doing the manual override in an indirect form so that the expected date will change and that will show up as a manual override. So at any point in time, you have removed one invoice from one day and put it on another day, by your own human judgment. That’s the option that we have.

[34:03] Amber Thompson:

Any other question?

[34:08] Audience:

Carol Stewart, so forecasting AP, I tell all 11 companies, what they can spend, and then they spend it and sometimes they sort of through our payment processors. So there’s the next day. Sometimes they write checks, they mail them in the regular mail, sometimes they FedEx checks. So in order to accurately forecast AP, since I said it and I know what’s going out the door. Is there a way to build that in? Does your model take into account that because I’m in excel and I’m continually working on when things are going to clear them off?

[34:52] Amber Thompson:

So you’re asking about whether we are capable to take total as into account or you’re asking about specifically, the delay that you might see between sending a check and having that payment actually apply.

[35:09] Audience:

Yeah, so I’m looking for things that are clearing. And how to know the time lag is clearing.

[35:15] Amber Thompson:

Yeah, we will go through an entire process of your payable system and how you set that up. And we are creating rules and models to be able to apply to your process.

[35:33] Speaker:

So let me answer that as well. So when the system does the forecast, it does not just take the invoice, the invoice date into account, but it also takes to account system behavior, like you know, hey, there’s a check or invoice which is supposed to be paid on this day. But the bank in a historical way shows that the actual payment is gone from your bank after three days or four days. So the AI-algorithm understands the behavior of the time lag to the actual payment date, according to your system and the effect in the bank. Okay, so it understands that and then builds the forecast. So the forecast is not just about on which day the invoices are marked as payable, but on the day that is actually expected to get out of your bank. Okay, the focus is done at that level. However, if you still want to override that, that’s where this manual adjustment option comes in.

[36:33] Audience:

Okay, thank you. Okay.

[36:37] Amber Thompson:

Any other questions?

[37:01] Audience:

In the instance where you don’t have certain debt in your ERP system, is it reflected in accounts payable, does it pick that up? So you’ve got six month, you’ve got one month, three months, six months? How far back is it looking? Or how far back will it take information to forecast handle it.

[37:30] Speaker:

Taking three years of data into consideration, so right when we are setting up, we take three years of your bank data which takes years of your data from your ERP from four different categories, invoices, receivables, and everything. And if there are any other significant cash flow categories, like if you have a very high variable contact, employees or something like that, we take all of that. And from these three years of data we are trying to credit, of course, at any point in time, the more data we have, the smarter the system becomes. So, it’ll start evolving. But if your seasonality is affected in the last few years of data, then it should be counted.

[38:12] Amber Thompson:

So, I’m going to bring us back to the spreadsheet for the biggest limitation points that were brought up earlier, and kind of go over how the HighRadius solution addresses these points. So it’s no longer difficult to collaborate across teams because HighRadius supplies you with one cloud solution where everyone has access. We also are no longer limited to simple formulas. We use it the AI functionality and the AI forecasting to be able to provide you with a smoother forecast. We are also addressing human error as well as that manual and time-consuming process by automatically pulling data from your bank, your ERP systems and any other appropriate data sources and creating a forecast from that before any users even touch the workout. And from here, I will actually let Paul go back into the benefits that he sees coming out of using HighRadius.

[39:38] Paul Watters:

So really, as we look to partner in and see what we’re able to produce out of this cash forecasting cloud, we’re obviously looking to get a much greater level of accuracy at a customer level. So the moment we have that across our portfolio level, but we want to see what the models are telling us to the customer level as well because that’s, of course, going to drive the accuracy of that forecast. We want to optimize the capital structure of our business. We don’t want idle cash sitting around, nor do we want to be calling off debt facilities wider than what we otherwise would. We want a better understanding of our past cash flow trends. You know, we talk about cash flows being important. And I suspect there’s a lot of us in this room that doesn’t actually measure our forecast accuracy.

Now, if it’s important, we should be measuring it. And if we’re not, then we’re not going to drive for better outcomes. So this system helps us look back over time, and also to be able to say to others that we’re reporting to them, and we provide that information to that. This is not a subjective assessment. This is based on the data, we can look back over the previous forecasts. And typically, we might be 97% accurate. So we can talk about the 3% of the margins. But what we shouldn’t be talking about is 10 and 20% differences, because the data tells us historically, that that’s not what happens inside. I think it takes some of the emotion out of it. If it’s not my forecast, somebody else’s forecast from international business, or from one of the other domestic businesses, it increases our credibility with those that we report to if we can be more accurate in this space, and I think that’s one of the key advantages. from a purely personal point of view, standards more objective, as I mentioned, no longer subjective doesn’t depend on how somebody’s feeling in the morning about what that cash forecast might actually deliver. And so that is really the last slide. So if there are any more questions about what I presented. How much time do we have? Any questions?

[41:54] Audience:

How are you managing the integration with Oracle and HighRadius through API, is it file transfers, is that a smooth process or challenging process.

[42:08] Paul Watters:

Each file transfers and so of course, getting access to the data and lights from our point of view is probably one of the challenges if you have a good recording tool of the system already probably going to be able to do that relatively. So, it’s file transfer, pretty reasonable thing was from the bank. And then, you know, once you have that ball built out of the article, then that, of course, makes it easier. It’s coming in overnight, and then the whole process is automated. So that’s what we’re working on now.

[42:35] Audience:

So to actually get the data out of Oracle, you had to hire an external party to produce it for you or something that you just export?

[42:43] Paul Watters:

So HighRadius has data extractors that can be used for this purpose. And so, where we pulled out data manually ourselves through various reporting tools at the moment, down the track, we will use one of their tools that allow us to connect into Oracle and bring out the data that’s needed.

[43:05] Amber Thompson:

And bouncing off of that answer. Hex is a program that we can use with SAP or Oracle. We have been using for quite a while now and our standard integrated receivables item. And so, it basically merges our client ERP system with our system pretty seamlessly. Any other question?

[43:41] Audience:

I just had a question that cash forecasting accuracy kind of where you’re at now and where you hope to be next year?

[43:51] Paul Watters:

So our forecasting accuracy is reasonably good. start wearing the low 90s at the moment, however, what you do find is that you also have situations where you might be in the lower 80s as well. And so, you know, as we deal with one of our largest customers makes up around 15% of our total revenue within that business, the payment behavior of that customer is pretty close to my weekly forecasts. Because, you know, of course, if that customer doesn’t pay this week and pays next week, then that’s something that I can’t see within the portfolio. And I’m, I’m regularly asked by our corporate clearance from the forecast by x. Why was that? And the truth is, I don’t know. I don’t know because I’m not proved to prepare a forecast by looking at 34,000 customers and deciding individually which ones of those are going to pay me at one point. So that’s not the way it works. And typically, you can have a look at some of your largest customers. The first is that it’s really difficult to decide and being able to look at forecast accuracy over time, and being able to drill down deeper into that into your top 10 customers and see some of those springs is a value proposition of the product to be able to get to some valid answers. The question is why?

[45:26] Audience:

As far as the drill-down capability, can you see when new data is imported? What caused the change in the forecast down the road?

[45:38] Amber Thompson:

Can you repeat the question? Yeah.

[45:39] Audience:

So if I come in today and I say you know, the end of the month, I’ll have $50 million, and then you guys do a data upload overnight. The next day I see open at $51 million at the end of the month. You can drill in to see what that million-dollar is from?

[45:51] Speaker:

One of the things I was talking about comes into the picture. So, in one of the grids that Amber was showing, you see the different forecast values happening on different dates. Like last Monday there was a forecast show, let’s say 11 million. And this Monday again, there’s a forecast, which says a million or something like that. And at each of these levels, you will have a drill down to show what are the invoices that were thought would come on that day. So when you do a comparison between these two, then you know that, hey, there’s one invoice, which was not part of last week, and there’s another one in this part of this week, or maybe one of them is missing, right? Either because the systems out it will get delayed or work.

[46:38] Amber Thompson:

And it looks like we have about two minutes left. If anyone does have a question that comes to mind later, please stop by our Treasury booth on the field. We’d be happy to answer anything else. And anytime this week, are there any other questions?

[46:52] Speaker:

You have detail demos over there.

[46:59] Amber Thompson:

Yeah. We’ll show you a demo that’s a little bit different from what we’re doing right now. So it’ll be both sides of the coin. Okay. Thank you, everyone. Thank you.

[0:00] Anchor: Thank you for joining today’s session on challenges and expected benefits with AI and MercuryMarine. We have with us, Paul Watters and Amber Thompson. Paul is the director of Worldwide Credit and Treasury of MercuryMarine. Paul moved to the US in 2015 to resume his current role as director of Worldwide Credit and Treasury for MercuryMarine. And we have Amber Thompson with us, Treasury consultant and HighRadius, Paul and Amber, the stage is all yours. Yeah. [0:44] Anchor: We also have a signboard with us. She’s going to circulate a signboard. Please write down your name and email addresses. It will help us to give you the presentation slides on email and also help you gain one credit for CTP recertification. So yeah, please don’t forget that. [1:07] Paul Watters: Good morning, everyone. So, my name is Paul Watters Hi, I’m employed by MercuryMarine, I’ve worked in their business for about 15 years, doing a number of things and opportunity kind of move in 2015 for the US and take up my existing role. So today what I want to talk to you about is an automation guide. I’m going to be talking a little bit about cash…

What you'll learn

  • Understand how traditional cash forecasting is done using complex excel formulas
  • Discover how AI transforms a treasurer’s workday and how it empowers decision making

There’s no time like the present

Get a Demo of Cash Forecasting Cloud for Your Business

Learn More

Request a demo

HighRadius Cash Forecasting Cloud – an advanced forecasting system – leverages the proven RivanaTM Artificial Intelligence (AI) platform to provide the most accurate cash flow forecasts – right from a ledger account level and rolling up to the organizational level. Delivered as a Software as a Service (SaaS), the solution seamlessly integrates with your company’s ERPs, accounting systems, banks and order management systems. Multiple AI and Machine Learning algorithms process datasets including bank statement inflows/outflows, sales orders/customers invoices, purchase orders/vendor invoices and expense reimbursements for comprehensive as well as accurate cash flow forecasts. The closed-loop, machine learning feedback system ensures that the forecast models become more accurate with time.