00:00
Patrick:
Welcome to the Mining Your Business podcast. The show all about process mining, data science, and advanced business analytics. I am Patrick and with me as always my colleague Jakub. Hey, Jakub.
00:12
Jakub:
Hey, Patrick.
00:13
Patrick:
Today's episode is all about the order to cash process, process mining aspect, KPIs, use cases, all that and more coming up next.
00:33
Jakub:
Hey, Patrick. So thank you very much for the introduction, as usual. So today we are really talking about order to cash process, which I'm pretty excited about, especially because it's been a while for me personally since I implemented it myself. But I know you did it implement with one of our customers quite recently, so I'll probably give the word to you right now, so Patrick, you tell me.
00:59
Patrick:
Okay. Well, yeah. Like I already mentioned, today's topic is the order to cash process. So what is it that we're actually talking about when we say order to cash you as a business? I mean, we could start with our coffee bean analogy if you like.
01:14
Jakub:
Oh, we do love coffee. And since, ah, it's early in this morning, I actually do sip my coffee right now so I can really tell you how this coffee bean goes from being a coffee bean into into my throat.
01:26
Patrick:
Yeah, that's, that's quite a journey. So when you want to sell your coffee beans, say you have a warehouse full of coffee beans, and you would like to sell them to customers who are willing to pay. You will get a order of some sort of this in this many coffee beans. This can also start early in the process when a potential buyer would look for a quotation, How many coffee beans can I get for this and this much money?
01:54
Jakub:
Right.
01:55
Patrick:
So then you have your quotation that gets to an actual sales order. When the when the customer says, yes, I would like to buy this, right? The order goes in and you all obviously now have to fulfill this order. You obviously need to do your stock checks. You need to do your credit limit checks, your delivery checks, and just make sure that you can get your customer his requested coffee beans, right? So once you have done all your checks, you will then have a transportation method of your choice, depending obviously where your customer is, package it up as much as you like and then have your goods movement leading your coffee beans away from your warehouse and on route to the customer.
02:39
Jakub:
So when I listen to you, it actually seems that there are a lot of dependencies on this and this order to cash processes. Once you get the order, basically it's not only on your, let's say, billing department or you know, on taking care of the ordering itself, but there is a lot of dependencies going into the plant management and then into producing it in the factory and everything.
03:10
Patrick:
Oh, yes. There's a massive interdependence between departments. And it's also very important for you to actually have these departments talking with each other within this process. Imagine that your goods leave your warehouse and nobody in the billing department knows, so they don't know when to send out the invoice. That's seven days or something that can happen where you just don't send out the invoice. You're seven days late on your cash and on your payment terms. Because nobody notified you that the goods actually left the warehouse.
03:40
Jakub:
Yeah, the deliveries itself, that's one big mess usually. Yes. There are just so many places where anything can get stuck and it doesn't even, like, need to get stuck technically, but only like on the paper that somebody is not aware of something being already received.
03:59
Patrick:
Absolutely. Now, I obviously already mentioned it, but once your deliveries are out, you obviously have to give your customer an invoice for the goods that he or she bought. Right. So that's another big task that is more in the financial side and the starting to get into the accounts receivable side. The invoice has gone out and you are now in a period of your materials being gone, your coffee beans are sent out, you sent the invoice, and now you need to wait for your customer to pay the bills. And that is obviously a whole another process itself, the accounts receivable section, where you have to record the payments that the customers make, that your activities, should they not pay, late payments and cash discounts that were taken, but they weren't eligible and things like that. That's a whole another process that goes into from the invoicing step all the way to accounts being settled
04:55
Jakub:
So just to recap on that. It seems to me that the order to cash process in general, we can divide it in a couple of phases. Like the first one being the pre sales activities that as you mentioned, there are some quotations. You might have contracts with your customers, you might have some scheduling agreements, meaning that your customer is expecting to be delivered, let's say one ton of coffee each month. So month after month, you are creating this delivery. So your customer and billing him each month. So that's the pre sales activities. Then the next one is actually the most important for us in this phase and that's the order processing. So what's going on within your sales department, how you handle all of the, the sales orders, so that everybody gets notified, everybody knows what's to do, you know where to send it, who to send it, who should send it and so on. The next phase is the order fulfillment and that's the delivery itself usually. So once the product is ready, you need to make sure that it's shipped from place A to place B in time, that the goods are issued and sent to your customer, and make sure that the transfer itself goes smoothly and is delivered on time. And that's one of the most interesting use cases in order to cash process. And that is the on time delivery, obviously.
06:31
Patrick:
Yeah, of course.
06:32
Jakub:
And then the last phase is the billing. And I actually was working on a process which was some kind of a, I would say reverse order to cash where instead of looking at the sales order, which I'm sure we will get into a minute, I was actually looking as my leading element as the billing. So the invoice, which was a little interesting because then you are actually looking more on the accounting side of the order to cash, but it was still not accounts receivable. So definitely an interesting point of view. And that brings me again to the general just process mining strength that it doesn't really matter what it just, you know, you can pick any item that you are interested in and then look at how does it look from its perspective. Super interesting.
07:24
Patrick:
Yes, of course. And like you mentioned, you can segment this whole process into various sub processes. Sometimes even helpful depending on your scope. Right. Because as we said, a lot of activities and a lot of different types of activities from different departments ar all in interplay here. So getting a clear view about this process can be a bit tricky.
07:48
Jakub:
Right? If we go into the technical details, what does it actually mean and what do we do if we are looking at the order to cash process.
08:01
Patrick:
So in your traditional order to cash process, you would want to track the activities of a sales order item Right. So this is a line item in whatever your customer has requested. In this case, it's our coffee bean. So that would be our sales order item that we track along the process from quotation to sales order creation, delivery, shipment, invoicing, and payment. Right. So what are some of the things that we can track? This is obviously important. We would obviously like to have some sort of entrance point into when the whole process starts. This can either be the creation of the quotation and then to the creation of the sales order. So you can see how long did it take for my quotation to end up becoming a sales order. When once you have your sales order, obviously it doesn't just stop there. There's multiple changes that can happen to your sales order. You need to go through a whole bunch of credit checks, delivery checks, stock checks to make sure you can actually deliver in the requested time period and so on and so forth. So there's a whole bunch of things that can change, be it the amount that the customer wants or that the price that the of the actual item that the customer wants. All of things can change in this in this part of the stage.
09:26
Jakub:
Yeah. The changes itself are the tricky part because those are usually the ones that are causing the delays down the road. So imagine that you decide to change all, let's say the quantity that is supposed be delivered to your customer by the time that you've already created your delivery documents. And the delivery is actually already being shipped away. And that's something that shouldn't really happen because then you are probably causing a lot of effort, unnecessary effort down the line. And so it should be prevented and actually with process mining and Celonis, we can easily track those occurrences down.
10:16
Patrick:
Yes, for sure. In the process that I implemented recently, it was the case that when one specific change happened and the change was when the vendor would not update the materials that were being sold on time. So sometimes they would just sell things that they no longer could sell because the article just didn't exist anymore. Somebody just stopped producing it. Hence they could not sell this. Right? So they had an outdated list of materials to be sold. A customer bought one and obviously they couldn't deliver. So obviously they needed to have the material changed. They needed to now update the schedules, the price, the quantity and all that sort of thing. And it took from a standard process of 20 days. It took 40 days in total, massive amounts of time.
11:06
Jakub:
The question regarding this observation, was it something that they already knew that is very likely existing in their process? Or was it actually discovered after implementing the process?
11:20
Patrick:
It was interesting. They knew that changes that involve material creation took a long time, but Celonis allowed them to really view what the root causes of those changes are, because you can obviously track what change initiated, what change in your process. So what you could really track down, hey, with all with like half of our material creations, it causes - hey we have an outdated list of materials that we're selling. So that was an interesting thing that they wanted to look at.
11:53
Jakub:
So that is really interesting. What else do we have there? Do you still remember in any other use case that actually you did for for the last customer? And that was actually interesting for them. As well, some discovery?
12:06
Patrick:
Yeah, for them it was interesting, just to kind of view it's the returns because obviously you can send out your goods, but your customer decides, hey, this is damaged or it didn't get there on time or in some fashion or form that they are not happy with and they would like to send it back. That can obviously happen. And when they send it back, you obviously you have to reimburse them or if they have already paid and things like that. So that whole process can just extend the order to cash by several, several days, months even, depending on when the customer decides to send back the goods and they wanted to know what's what's the overall picture of our returns. Are we talking about a specific customer that always sends things back? Is it because at during delivery with one of our transportation companies that we hire, is there a bigger rate of returns in some regard or can we figure out the cause of these returns and try and minimize them so we don't waste that much? One, manual activities dealing with them and also just having to give back the money that the customer has already paid?
13:16
Jakub:
Yeah, I know that. Sometimes when the reality strikes when a customer sees the process for the first time, it's like, wow, I don't believe it. And the first the first reaction is usually that exact disbelief. And once they actually start looking at that case by case, which in this case is actually represented by the sales order and they start checking their ERP system, in this case, very likely SAP, they actually start believing it and start seeing some previously unnoticed occurrences of some undesired behaviour. It's amazing how quickly it goes from disbelief to disenchantment.
14:04
Patrick:
Oh, yeah. Oh, yeah. We we set that up and we had our first a kind of view of slowness and the things that we implemented. And I think one of the sponsors of the whole project really scratched his head at some of the findings that we had, you know, where we just they just ran out of stock at some point. And just didn't refill for seven days. And the customer was waiting for no reason whatsoever other than the fact that they just completely missed it. Right. And this was a noticeable occurrence so there are definitely metrics that you have in mind. But once you get a clear picture of the as is, you can really start making value driven decisions in that way.
14:50
Jakub:
Now, we are now in the midst with one of our other customers of implementation of a process. It's not really order to cash, it's more like order to make. And that's exactly a customer that is producing material that is quite time demanding. So it's not that they can just ship it the next day. It takes some time for preparing and for creating it as well. And for that reason, they need to be sure that they're letting all the make all the material actually is maintained well enough so that they have enough of time actually producing the material and then shipping it to the customer so that they don't actually promise customer, yeah sure, we can deliver it in a week and then they find out that they actually first contact their vendors to buy all the necessary parts. And, you know, it can be it could be an automotive, it's like if you are, if you are about to deliver 20,000 pieces or something of a piece that's 20,000 cars, you really need to have your supply chain maintained to such a detail that you can actually know that you will deliver it in time.
16:01
Patrick:
Oh yeah. I mean and that's if you an automotive you have your several products that are all almost identical in a way. Right. But other companies, for example, they will prioritize some customers over others. They are higher profile. They have more business with the, with the company and so on and so forth. So they will have a higher prioritization on the materials. Right. So all of a sudden, reservations on some materials that you needed for your sales order items are now just stolen by other ones. And so this is obviously a place where you need to look at your stock and not just say, oh, we have enough stock, this will be fine. What are the odds that some other sales order item for this particular material might come in and we might not have enough? You know, these are trends that you can look at. They happen for specific materials or for specific customers that you have. And you can really see trends forming where you can make better decisions about how to stock your warehouses. How to prioritize customers and based on past behavior, how you can augment your process in the future.
17:12
Jakub:
And just listening to these use cases and to what you're saying, if you are a big company, you need to process mining, really. We don't even really need to do any any promotion of that. Just knowing what is going on behind the curtain and you know that you need to start tracking what you're actually doing. Because being in charge for maintaining a huge company of thousands of employees and so many stuff that is going on in there. That takes a lot of efforts and you better know what you're doing.
17:49
Patrick:
Absolutely. And it might not even be a big, big hurdle or a big pain points in your process, right? It's not that, hey, we are all of a sudden billing 20 days too late or something like that. It can be tiny things along this massive process that can all just add up, right? If you can reduce the amount of manual activities, the throughput time of your process, make them a little smarter, make them a little bit more efficient, you can save a lot of cash, manual labor, headaches and so on and so forth.
18:26
Jakub:
Yeah, that's also very, very interesting. I know that sometimes when we are implementing use cases for the process mining. What we do is that we actually put some number on each activity meaning that we can track it from the perspective on as you are saying that the manpower that is needed to create such an activity, being manual activity. So if you are creating sales orders, you can create them either automatically. So it's some kind of a trigger by the system or you can create them manually. So first of all, if you do that manually, there is some room for error always because we are just humans and we tend to do mistakes, especially when we are typing some very long number into the into the ERP system and what we can do then is that we just put a number on.
19:23
Jakub:
First of all, how long does it take? So let's say that you if you are creating your sales order manually, it can take you let's say 5 minutes and then you see, oh man, I'm actually creating 40% of my sales orders manually. And if you have 500,000 sales orders, that's going to be like 250,000 or 225,000 or something of sales orders being created manually each year. Let's say you put a number on in 5 minutes and this is some mathematics, it's a bit beyond my comprehension but you get my point. It takes a lot of time that is unnecessary and is very easily or can be very easily automated because usually your sales orders come from some steps before. And once there is this step in place, you can easily automate and so these are the pain points that then on you can put some number on it and you can either you can just go to your boss and say, like, look, if we automate this specific activity down the line, we cannot almost say, let's say those are a lot of hours of someone work or tens of hours, but we can also put not only time on it, but also a price on it. Because the person who is performing this is needs to be paid. Right. And if you can save not only this time, but you can also say obviously some money down the line by not paying this person, but also making the process much faster.
21:04
Patrick:
Oh, yeah. And I mean, so you've kind of hit the nail on the head there. How do we measure the health, the efficiency of the O2C process, right. So we have KPIs that we obviously have, but we obviously need to then figure out, well, how much is this costing you or how much is this slowing you down? How long How much time could you optimize? How how quickly could you get your cash? These are kind of the metrics that we need to talk about and kind of frame our analysis and our knowledge to kind of get some of the required results. Right. So if it is like you mentioned, we want to have higher automation rates, right? The manual activities, the people doing them would probably want to do more important stuff in their day to day rather than creating things manually. So this is already one of the better things that we can look at automation. That's a good one.
22:05
Jakub:
Yeah. The theme is usually very similar in any process mining or any process. There are a couple of things, high level things that you can always look at, as you mentioned, the automation, right? It's obviously one of the easier ones as each activity happens in the system. So what we are really doing is that we are looking at who perform this activity. And so to pick up on the example I was talking about earlier, that was our sales order that was created by someone. And because we have this information by whom it was created, we can obviously track it down in the system and check for the user type of such a person. Whether it was actually a real person, then we would have a certain user type or whether it was automated by the system. Then we have another user type from this. We can just subtract the ratio of automation and manual activities and just this is one of the strengths of Celonis is that it's almost seamless and it doesn't even take much effort. So in every process there are a couple of things that are always first to shoot for, automation being one of them.
23:20
Patrick:
Just to get on that point about automation, two things to note that actually we don't really see names in our, in our analysis that most of the time, usually companies prefer to have their, their employees activities anonymous. You know, it is a little big brother in that way that you don't want to obviously micromanage every little thing that your employee does in their day to day. So when we reference the whom, we are speaking about some anonymous person in a company that's obviously important, you know, privacy and that sort of thing.
23:54
Jakub:
There usually are four letters that come up in the first scoping workshop and that's GDPR, but the good thing, if your business is based in Switzerland, then you probably don't care about this and we can go even with names.
24:12
Patrick:
Yeah, that's true. That's true. And the other thing that I wanted to mention was that we can obviously look at automation rates and the user types. This was actually a point that was brought up in the last order to cash because we were looking at the automation rates and we saw that the user tables were just not properly kept.
24:32
Patrick:
So we didn't have a real knowledge about who was a machine and who was who was an actual person. So that was another task that they now want to tackle is just kind of cleaning some of their data out and amending it.
24:46
Jakub:
Let's see the master data. I would like to also say that you hit another nail right here. The master data can be tricky as well because maintaining master data is usually a manual activity. And we know what manual activity brings with. So I know from experience that what sometimes can happen or shouldn't happen, but what can happen is that you essentially create let's say, two similar customers or two similar vendors with different numbers, but almost identical or in technical technically they are identical. And then what can happen is that the one time you bill one vendor and the other time the other vendor, while in the reality it would be the same kind of vendor. And these are some inefficiencies that can cost you a lot of headaches down the road.
25:38
Patrick:
Oh, yeah. Oh, yeah. So sorry. I was breaking up your flow, you were about to mention other KPIs in the O2C.
25:45
Jakub:
Sure I had there another one and again that's easier to do once you start implementing process mining and those are the throughput times because the way the process mining works and the way you are actually creating your activity log, you have the timestamps of when the activity is actually performed, when is it happening and it's very straightforward and relatively easy to compare the differences between the different steps in your workflow. So you can just try to head go and check, check how long it takes in creation of your sales order until your goods is shipped or creation of sales order until it's actually built. And then you can see where are the bottlenecks. You can just drill down them on different kind of engines, check where it could be actually happening and see the most problematic ones, where the activity itself takes a very long time until some other action is performed. And you can narrow it down to like 5% of worst performers and just go from there.
27:02
Patrick:
That's interesting. We also have the amount of changes that a process has, as we already mentioned, beforehand, sales orders and delivery documents, accounting documents and all these things can be changed. And it is because of these changes that it's not a easy process to to handle changes in price and quantity. You can obviously change your delivery times and they can have a lot of knock on effects. Right. So we would like to track the changes when they occur. At what point actions based on these changes occurred and how how each action kind of cascades throughout the process. So you could identify specific changes that have a bigger impact and some that are not only meaningless like in a throughput time sense, but also just necessary checks.
28:00
Jakub:
Other use case that comes into my mind and I already mentioned it, it's one of the first ones that are usually implemented and that just a simple overview of On-Time delivery. And that's I already mentioned that the time information is quite easily obtainable and so what we can do in Celonis is that we are we really look at each and every sales order and we are looking at when it was actually whether it was delivered on-time or not and we can even play with it. So we can look at the early deliveries, we can look at the late deliveries and we can even create some kind of offers in there so that the end user eventually have a power to, you know, create some rules when it's still considered to very early, when it's still considered to be late and so on, so forth. So you just give them a very powerful tool in their hands when they really can go down into the deliveries and use this process mining tool as reporting tool that they then go to their customers and actually their departments and tell them: "so this, this why we are not performing well and because of these late deliveries, we might be losing one of our core customers" because if you deliver late something that your customer is expecting to be delivered by the date that's causing a lot of problems.
29:41
Patrick:
Oh yeah, it's actually one of my favourite use cases because we can kind of do a predictive analysis. We can say based on your past data, this material or this root or this customer, it is predominantly late. So we can kind of implement some sort of early warning system that as soon as the sales order comes in, based on past history, this has a very high likelihood or low likelihood of being delivered too late. So you can already kind of change some of your actions based on the specific details of that specific sales order to kind of prevent late deliveries.
30:33
Jakub:
That's that's beautiful. And then you can leverage all the strength that's in Celonis. Also you can just go ahead and create those warnings. So leveraging the action engine in Celonis and just send the emails to the people who are responsible and alert them on some undesired behavior that are either good or is already happening. And funny thing is that this prediction that you were mentioning, but you can also do is improve it in time because then what the user can do is just track whenever the alert was actually justifiable and when not. And just based on this input, based on this feedback, we can just go on and improve your algorithm, what is in the back end and on the go improve the whole process.
31:29
Patrick:
Oh, yeah. You will start to see that the metrics that you use to measure your process health, like these late deliveries will hopefully go from the red into the green over time so you can watch your business, your process become better in real time.
31:45
Jakub:
Yeah. Another use case that I remember implementing with one of our customers was a very simple overview of an open sales orders. So what can happen and will happen in time that some of the sales orders are not converted, meaning that they were not to delivered and not billed or they have been delivered but not have been billed and so forth. Usually that's quite a standard behavior because obviously once you create your sales order and let's say ordered delivery, it still takes some time to get all the necessary information to your billing department and create that invoice eventually. But sometimes things just get forgotten for whatever reason, and you can then track down in your system sales orders that have been opened even for years.
32:42
Patrick:
Can you kind of give me an idea of why that would happen?
32:47
Jakub:
Well, I don't remember exactly, but I do remember that once we created this report with one of our customers, we did have their sales orders of not really insignificant amounts that have been open for like five to ten years even the reasons could be I mean, eventually it could have been just canceled somehow, but not tracked in the system. Or it could have even been already paid for. But the invoice got lost somewhere. And so there are some reasons, but it's actually something that you have to then go on and track it down in your system manually. What you can do in such a tool like Celonis, is that you can get the overview of such occurrences. But the reason why is that happening? You could obviously check your process and try to find some better why it could be happening, but most likely it's some human error.
33:50
Patrick:
Yeah. So incomplete deliveries don't mean that they weren't paid for or weren't actually delivered or where invoices weren't announced. They were. You just didn't make the connection in your system, right?
34:05
Jakub:
Yeah. True.
34:07
Patrick:
Yeah. All right. So a few more a few more use cases here we have the segregation of duties. I know you did this some time ago with one of our clients, right?
34:22
Jakub:
Exactly. So segregation of duties, it's something you can implement again, almost in any process, because what do you do in segregation of duties is that you try to prevent some undesired behavior. And in this case, that is one user or forming two activities at the same time that he or she shouldn't. So this could be, for instance, creating a sales order and then improving it or I'm running out of ideas right here. But generally speaking, it's something that shouldn't happen at the same time. So if somebody does one or performs one activity, then the other the follow up, that always needs to happen should be done by someone else. And that is mainly, there are a couple of reasons, but one sticks in my head, and that is to prevent some potential frauds.
35:25
Patrick:
Right. So you're saying some users cannot remove the credit blocks, the delivery blocks and the invoice blocks all at the same time and just have this.
35:34
Jakub:
Yeah. Yeah. So it's compliance, fraud detection and a couple of other reasons that can be deducted from the segregation of duties and just to preserve the system and make sure that everything's working as it should work.
35:52
Patrick:
Oh, yeah. And this is obviously very important. If you need to have compliance checks on the regular as some businesses are more prone to having them than not. You obviously want to have transparency and, you know, obviously abide by all the compliance rules.
36:07
Jakub:
Auditors do like this report.
36:10
Patrick:
They do. All right. So I'm glad you already mentioned it. Action engine, right? It's all good and dandy that we can look at the transparent process as is have our analyzes, have our KPIs and things like that. But what a really good strength is in Celonis is the action engine. So from there, we can have our signals that go out based on the analysis that we create. So say, hey, this is a sales order that is highly likely of being late delivered so a signal goes out to some user informing them of an action that they could potentially do or they could help prevent something from happening. So this is where we take the as is data and in real time affect people that need to be affected to comply with new KPIs or prevent late payments and so on and so forth.
37:12
Jakub:
Yeah, not much that I have to add to this actually.
37:19
Patrick:
Yeah. So this is a very big strength that I think we still have because a lot of customers don't know that this is viable or that this is even a thing that they can actually implement some groundbreaking, not groundbreaking, but like effective change in their process in real time. Right. It's not just a bi tool. We can actually effect positive change and have an influence on the process, as is through this action engine. And I think we need to obviously make customers aware of this and you listener out there that we can do this, that we can help you in a way, which is the ACTION ENGINE, coming soon to your processes. Now, these are obviously good tools. And I think they have some some decent to some good value for sure.
38:13
Jakub:
And that's why I keep saying that process mining project is not a one time event, but it's an ongoing incremental improvement. So once you do the first implementation, it just opens up so many gates that you can go through and start focusing on. And so the main job for us is to try to maintain the focus of the people on the things that matter and try to go slowly with this so that you identify a couple of use cases and try to go and dig deeper of those rather than just go many, many directions at the same time and completely lose the focus.
38:58
Patrick:
I mean, that's why we often talk about the process mining journey. because it is quite a it is quite a journey.
39:03
Jakub:
It is. Not the only for customers, but also for us.
39:10
Patrick:
For sure.
39:11
Jakub:
Right. Patrick, I think that's probably a wrap for today. Hopefully our listeners do have a bit better grasp on what is order to cash process. If you have any questions whatsoever, just let us know.
39:31
Patrick:
Absolutely. Well, well done. All right. That's a wrap. See you next week.
39:36
Jakub:
All right. Bye, Patrick.
39:38
Patrick:
Bye bye.