Good morning everybody and welcome to our webinar on Integrating Macroeconomic Indicators in your demand planning process. We are very excited to have you all here today and I can see in the chat that we have people joining from all over the world from Denmark, Switzerland to Singapore. It's super nice to have all of you here to discuss this very interesting topic today. So we are joining you from Denmark today so let me introduce you to who will guide you through the webinar today. We have three hosts today. Number one, we have a special guest from Grundfos joining us, Pea, who will guide you through how Grundfos is succeeding with integrating macroeconomic indicators in their demand planning process already today. And then here in Copenhagen we have Mike and me who will be your hosts for the first part of the webinar, guiding you through the concept of macroeconomic indicators and how to integrate that in your demand planning process. But before we start, we asked you all a question. How strongly do you believe that macroeconomic indicators influence your demand? And what we usually experience in organizations is that economic trends are acknowledged but rarely anchored in the demand planning process. Therefore, we would like to guide you to three topics today. First, we would like to discuss together with you how you can turn your macroeconomic data into actionable insights and real demand signals. Then, number two, we will guide you through how to use the macroeconomic indicators in your demand planning process and in your demand review meetings. And then thirdly, we will see how this is actually done in practice by companies already today when we hand over to Peer in Aarhus who will explain how Grundfors succeeded with this and how this is running in their everyday business already now. Cool. So let's start and I will guide you through the agenda and then we jump right into the topic because we have exciting 45 minutes ahead of us. Our agenda today is the following. We will start with Mike who will guide you through the agenda. We will start with Mike who will guide you through the agenda of macroeconomic indicators in your demand planning process followed by what is actually a macroeconomic a good macroeconomic indicator and which indicators are seen to be rather weak. And then we will jump into how does this look in your demand planning process, how to guide your demand review meeting in a way that you can actually make decisions based on these macroeconomic indicators. The last 20 minutes we will spend on the case insights from Grundfors where we will hand over to Peer. In the end, we will wrap up all together with the joint Q&A. So if you already have questions throughout the next 40 minutes, please put them in the chat and we will try to address them later on or come back to you. I'm super excited. How about you, Mike? I'm also very excited. I've been looking forward for today. Yes, then I'll hand over to you for the first topic. Thank you. So first of all, why at all consider including macroeconomic indicators into your demand planning process. So first of all, these indicators, you can use them as early warning systems for potential market shifts. So normally we see if there are a change in a trend, then it could impact your sales a few months after. Secondly, using them can improve the consensus and cross-functional alignment. So what I mean by that is that this consensus is normally driven by what are the assumptions behind the plan. And the assumptions, if they are based on the future outlook of the economy, then it can improve the common rationale of how that looks like and also improve then the alignment. And finally, we also see that including them, the indicators can also help reduce the long range forecast bias from your forecast. So that is why we believe including these indicators can be useful in demand planning. So now consider this scenario here. You have the previous years of sales and you need to commit to a forecast. And the forecast, that is one element of guiding you how to make the best supply chain decisions. Normally, such a forecast for your S&P horizon at least consists of three main components. You have the level. So what is the yearly sales you expect? How does the seasonal component look like? And what is the trend? And looking at this example here with the past sales history, it can be fairly easy to identify what is the level and what is the seasonal component. But in order to understand the trend and if you expect increase or decrease in sales, sometimes it helps to go a step back. Going a step back helps you easier to identify what is the overall trend. And these trends are very important for the S&P decisions. Because it is in the S&P decisions with the trends that helps you understand whether you need to invest in new capacity, whether you need to change the footprint or whether you need to consider production shifts. Understanding the underlying elements of what drives the trend is what actually helps you improve the S&P decisions. So one thing is the trend, but you need to understand the underlying elements. And you could say this trend that is actually driven by two main factors. There are the external factors, which are one of them is the macroeconomic indicators. And basically, these macroeconomic indicators, they shape what is the total market development. And then there are the internal factors. So that is basically what factors you can do that influence your share of the markets. So it's a combination of the external factors and the internal factors that derives what does the total trend look like. When we look at the total market development, that normally follows what is called business cycles. And these business cycles, they shape the overall or that describe how the overall economy is shaped. And these macroeconomic indicators, they are used to predicting these business cycles. And they have an impact on your sales and thereby also on your forecast. So if we look into these business cycles, then they consist of four phases. You have an expansion, that is normally when the economy grows. You have a peak. The peak turns into a recession. And when the recession turns, then it's a throw. So these four phases, that is the business cycles. But of course, the length of the cycle and the length of the phases varies from cycle to cycle. From a demand manager or S&P manager perspective and considering using macroeconomic indicators, it is important to understand, first of all, what business cycle phase are the markets in currently? What is the expected length of both the phase and the cycle? And also, is there difference in your product portfolio and across markets? Because of course, if you are a global company and you have a broad range of products, there will be different indicators impacting your sales across the globe. And then finally, what are then the macroeconomic indicators that actually affect your sales? And these indicators that affect your sale, that is, you could say, a part you need to consider of what does a good macroeconomic indicator look like? So maybe just to say, now we've talked about macroeconomic indicators, what is a macroeconomic indicator? So basically, there's a public available statistical numbers, and they're used to assess economic health of a country, a region, or population. They're normally used to guide decisions for governance, investors, and businesses. And that is where we see the change, because historically, to be used in a business context, it has either been used on a strategic level or how to support go-to-market strategies. And this is now where we see the change that is now also being included in the SOP process. And then it's important to consider that these indicators, they can act both as leading, lagging, and coincident indicators. Yeah? So there are some characteristics that you should consider when choosing an indicator. And these four here are very important. So first of all, there should be correlation between the indicator and your sales. So that basically means that they should evolve in the same direction, or at least they should have the changes at the same point in time. Secondly, then you have the leading... Yeah? Wait, Mike, sorry to interrupt. I'm just wondering, isn't there something about causality versus correlation in this? Yeah. Okay, you're right. Thank you, Marie. So one thing is correlation, but of course, causality is even more important. So when choosing which indicators are important, you need to have a hypothesis-driven approach and understand what can actually explain the development in your sales. So causality and correlation are important. Then, preferably, the indicator should be leading. So that basically means if the indicators are turning from a peak to a recession, what is then the time shift of months from the indicator signaling it until it hits your sales. Then there should be timeliness. So that means the indicators should be available when you need it. There are some indicators. They are first updated maybe one and a half months later than the point in time. So for example, indicators from January are updated here in March, and that will be too late for business decisions. And then finally, the indicator should be stable. So that means that there are some indicators that are revised frequently because maybe they are compound indicators. And every time these indicators are revised, that means you need to change how you use them in your modeling process. So these are the four indicators or the four characteristics you need to consider when selecting an indicator. And what we have seen working in business is that on the strong indicators, we kind of see a difference between how far up or downstream you are in the supply chain. So for B2C companies, then the indicators we see that have a strong correlation with sales are stuff like employment, private consumption, household growth, disposable income and investment. Whereas if you are further upstream in the supply chain or have a big export, then there are other indicators that are useful for you. So that could be industrial production, GDP exports, and there are also various specific sector output indicators that can be used. What we also have seen is that there are some financial indicators that, well, there might be a correlation, but not really any causality. So, for example, current account balances or external debt, that they are weak indicators of explaining what is the overall sales, at least for, I can say, a normal production company. Good. So that was a bit of the background of why we believe it's important and what a good macroeconomic indicator is. So now, Marie, maybe you'll help explain how to use them in your demand planning process. Yes. Thank you so much, Mike. The first part was already very insightful. Now let's dive into how can you actually use these macroeconomic indicators in your process. The first thing is looking at a best practice demand planning process. The reviewing macroeconomic indicators belongs in a business-oriented demand review meeting. So how do we actually do that? Many of you might be wondering, our demand review meeting is rather focused on handing over an aligned forecast to supply, which makes the conversation rather supply-driven than commercially driven. So how can you succeed in shifting your focus from supply-driven to commercially driven? And in our perspective, there's two main things in order to do that. The first thing is that you need a purposeful meeting for that, where you really center on this idea. And the second one is conversations. Let's start with number one. We see that macroeconomic indicators, they explain structural demand direction over the next 18 months. So they give you a context to your long-term forecast outlook. And therefore, our suggestion is that you run macroeconomic indicator reviews on a quarterly basis in a quarterly business review. How does that look like? Thinking about your standard process, you would run month one and month two of every quarter as a usual monthly demand review meeting, focused on your volume forecast and handing that over. Every third month, you would extend the agenda in order to make it really focused on the commercial outlook. You integrate the macroeconomic indicators and you have specific conversations around those. But which conversations are these actually? And therefore, I would like to dive into three conversations with you that we see are very important in order to make this meeting count and to take decisions based on the macroeconomic indicators. So what are these three conversations? The first one is that we strongly believe macroeconomic indicators are a very powerful tool in order to perform a bias and realism check. So you can ask yourself, are we biased by last year's economy? Are we, for example, coming out of a recession and our forecast is still very conservative and we're risking to miss out on the growth spurt? So we are not seizing an opportunity that's coming up. That's number one. Number two is decision implications. If we look at this macro signal and it might be contradictory to our forecast signal, what are the decision implications? So which decisions do we actually need to adjust based on this macro signal in order to derive the right actions? This could, for example, be very useful when you have a risk to assess. That's coming up. And then the third one is market versus company ambition. And this is really where you can seize the commercial opportunities so that you can imply this business orientation, get commercial in the driver's seat as well, and connect your business objectives with your planning outlook. And that's market versus company ambition. So the question to ask yourself here is, are we planning to grow faster or slower than the market? And if so, how do we want to do that? So what does this imply for us to do? I would like to take this opportunity to dive into number three a bit more in detail so that you can see how this example can actually look like in practice. For that, I brought you a little data example here so that you can understand how this conversation could be shaped or how you could phrase it. What we see here on our slide is our long-term outlook or our outlook over the next four or five quarters. And we have two data charts. The first one is our forecast index, which indicates how our forecast is developing in the long term. And the second one is the macro index representing the macroeconomic development over the next time. If you see the line here in the middle with the 100, that will be where you have no growth but stability. What we can actually see in our example here, and this is where the conversation becomes interesting, is that we have a forecast index which is slowly picking up. So it's suggesting that we're planning to grow. While our macro index, after a decline, is actually suggesting stagnation. So we have different directions of our two indexes. So what do we do with that? If we are planning to outperform the market, how do we do that? And who needs to be involved in order to make the right decisions to achieve this ambition that we have here and really synchronize our demand planning with our business objectives? And while this was a very first glimpse into how this could look in a practical example, now I would like to hand over to Per in order for you to get insights into how this is done in the Grundfors business on a daily basis or in the demand planning process today. With that, Per, I'm handing over to you. Thank you. Thank you, Marie. And then I will guide you through how we approach this in Grundfors and taking these indicators into our demand review process. Yes, and just a small introduction of me. I'm heading up the global forecasting team in Grundfors. And a bit looking back on my career, I've always been involved in SNOP and especially in the demand planning process has always been part of my professional career, both through my time in TULIP and now in my daily work at Grundfors. Looking a bit also to Grundfors, for those of you who should not be familiar, it all started back in 1945 in Beringbro in the middle of Jotland. We are roughly a bit more than 21,000 employees and are mainly owned by the Paul Du Jensen Foundation. We sell a bit more than 11 million units of different pump solutions every year and that equates to a revenue just below 5 billion euros. We heavily reinvest in our company and roughly reinvest 6.4% of our revenue. Looking at our organization setup, we are divided into four divisions and then we have a group function as well. And we are present in more than 100 countries around the world. When it comes to demand planning, we have a demand planning function sitting in each of the divisions who carry out the demand planning for each of the prospective divisions. And then we have a group function to support and develop the process overall. Then looking a bit into the macro and also why we kind of embarked on this journey, we came from a situation where we had statistical forecasts, like I think a lot of companies do. And then we were kind of faced with the question, what would actually happen if GDP was increased by X percentage point? How would that impact Grundfors sales? And that we didn't have a straight up answer for. But it kind of ignited the modeling of this and how could we then try to build some correlation between what we saw in the outside indicators and did that have any correlation with our sales? And we also saw that this will not only kind of bring clarity of what are the correlation, but also a bit more what are the lacks of these indicators? Is it something when we see development in macro indicators, will it then project onto our sales within one month, two months, or even further out? And then we can use it more actively when we are predicting and reviewing our sales. We started to formalize a kind of structural approach and having a project organization around it with a steerco where we both have represented from the division who was kind of initiating it. We were also part of it from group. And then we had to implement also on board to guide us through the process. Then we had a core team to look into all the details and do a lot of work. And I think here it's important to highlight that some mathematical power is definitely needed because it is rather complex exercise to do. So be sure that you staff up on these capabilities when you embark on this journey. We also had a reference group and this was especially useful for us in order to have people who know the business, who know the market, and also to have insight to the history of the company and what kind of development we've been through. Then we put some design criteria for the process that we wanted to have in place. First of all, the orange bricks, the market indicators, and that was important for us that they were publicly available and they were also available in a format that we could easily integrate them into our models. So they needed to be available in some electronic format that we could import. It's also quite important for us, represented by the blue areas, that it is leading indicators. We don't want to predict the historical sales, but it needs to kind of predict something in the future, as Mark also alluded to before that. And when it comes to the boxes to the right, we want also to go with what we call the adjoined macro signal. And that kind of means that we want it all to be combined into one number. We didn't want to end up with five or 10 different macro indicators with different signals. And then it kind of is up for the individual ones in the demand review meeting to kind of interpret which one has been more priority over the other. So we wanted to end up with one number kind of representing how do we see the outside world potentially impacting one for sales. And then we wanted to be looked into the demand review meeting in a kind of pragmatic and non-scientific way. Next steps I'm going to show you is a bit about the model creation. How did we kind of approach that? And we went into these five steps when we developed the model. The first step was kind of to do a visual inspection of our sales and see if there were any special events that we need to take out. That could be some kind of sudden impacts or whatever that kind of is an outlier and not to be included in the historical sales. Then the next step was to kind of go through the indicators that were selected and see which one when we do a visual kind of plot our sales. Again, the indicators, which one of them has a correlation just from a visual point of view. And after that, the one that we identified as having a visual correlation, then we start constructing the correlations and then also start to build the more specific coefficients and see how we correlate with the data. The fifth steps and also quite important for us were also after we created the model also to do some back testing to see how much does it actually explain of the development. And that is also important in order to build trust and confidence in the organizations that this is actually working. And it actually does explain some of the variation in our sales. It's also here that we also realized that some of them, we have a low accuracy and then that's of course a consideration of the event to continue with these indicators or not. Selecting the indicators is also a process on its own. And that's also where we kind of relied on the reference group, both with their insights into the market. What is relevant to take into consideration because there's a lot of micro indicators out there. And it's really a tricky part to find the right one that has some insightful and meaningful correlation with the sales. So this was also quite a detailed process for us to win through which one does actually make sense, but we want to follow through the process that we saw on the previous side and then do this kind of gross list. And then do the exercise market by market and product segment by product segment. We kind of started out with roughly 20 different indicators that we ran through the previous process. Then we kind of then identify which one to go with. Then it's how do you want to combine them into this one model who predict this one macro index. And just to illustrate a bit the complexity here. So this is an example where we have three different indicators and there are so many different ways you can combine. And do you want to go with a model that has only one variable or two or all three of them? And it's just adds up to multiple ways of combining these into one joint macro index signal. So here we also kind of establish a process and a program to validate these. And at the end of the day to come up with what is the best model to predict this joint macro index. But also to take into consideration that we don't want to build a too complex model. We want it also kind of to balance out the complexity with not bringing too many indicators into the model. But at the same time also achieving an acceptable accuracy for the future predictions. So after calculating the model and also calculating the indexes, then how to then use it. And also as Marie said, it needs to somehow go into the meetings. And of course, we calculated this joint macro indicators index. And it was important for us to keep it with only one number. So not having multiple index for different indicators. So that was kind of our outcome of the models. But also bearing in mind that macro indicators only explains something. It explains how the overall macroeconomic environment is doing and how this potential can relate to one for sales. However, there are also things that it does not capture. It does not capture if certain political legislation changes. It also doesn't say anything about how marketeers are developing. So it is more kind of an outside reference to how sales of the market are developing. And of course, there can be differences in how our sales are going into the future. So what we did and how we end up using it was kind of looking into what is our growth index when we talk about our future plan. And we also here combined the growth index to be also a bit more aggregated level. And then we kind of converted our growth index into errors in different color codings also to make it quite problematic. And then we kind of combined also the combined macro signal into these errors with the same color coding. This was also done so that we didn't see the value in showing an index if it should be 124 or 126. Because we can anyway not predict the growth that specific from markets, but more than turning it into is it a dark green representing a high increase from market? Is it a light green or is it more constant that we see with development? And then it's kind of the last line in the table. So does our development follow the same direction as what we see in the macro signal? And of course, in the quarters where we see deviating trends, then of course it sparks a dialogue and see other kind of assumptions why not follow the market. And I think that's really where the good dialogue starts in the demand reviews to try to these differences. We also follow up on the accuracy and also to give some validation to see how accurate is it. Are we macro in combined signal actually able to kind of forecast the way that our sales is going or what are the accuracies at a high level? So we also can build some trust or see if there's something we need to look into to improve this. Going to another use case, and that's where we kind of went a bit outside the demand planning meeting, review meeting. We also at a later stage decided that we want to embed these macroeconomic indicators directly into our AI forecast models in order to kind of give them as a good starting point and be able to predict as a good baseline for our divisions before we start to look into the demand plan and see if we should and where we can add more values on top of the statistical models. It has also in this process been quite keen to us that we only add in macroeconomic indicators that we kind of validated up front. So we don't want to kind of overload our AI models with all different kinds of macroeconomic indicators and then lead it up to the models to figure out which one correlate or which one do not correlate with our sales. So we have a bit strict gatekeeping at this point and I think that's quite important for us to keep it like that. So overall, we kind of have two use cases for where we kind of use macroeconomic indicators. The first one is that we use them as an outside-in validation in our forward-looking demand plan in our demand view meetings. So we have a bit of a different direction and we have a different direction and also realized that there were more indicators that kind of had an impact to our sales. And that's why the reason why we went into this more joint macro indicator signals that we kind of review in our monthly meetings. Then secondly, it also provides input to our AI models and that is now embedded in the models. So we run this on a quarterly basis, meaning that we also it's on a quarterly basis that we update our macro models and get the newest outlook and the newest indicators. And then we calculate the indexes so they are up to date again. This was how we did it, how we do it at the Grundfos using macroeconomical indicators. And I think now I'll hand it back to you, Mike and Marie again. Thank you, Per. That was super interesting. There's a lot of insight and we can see that there's many questions coming in. I would like to just pose a question straight away to you, Per. We got a question in the chat, which is asking, how long does it did it actually take you to implement the macroeconomic indicator model and process in your company from kickoff to finish? That's a good. I think it took three months. I don't know if you recall it, Mike, but it is a couple of years ago. But I think roughly that was the time we spent. Yes, I think what we focused on was identifying some pilot markets, right, to test it out. And then I think we used three months, right, to build the structure and to start using it. And then I know afterwards you have been rolling it out in other business units afterwards, as I understand it. Right, Per? Yeah, we've been calculating it for another business unit and are about to embark on also deploying it. Yeah. So three months, you could say, from the kickoff and until it was ready for, you can say, the pilot markets. Yeah. Cool. Thank you for answering that question. I would actually like to dive into another question, which is towards you, Mike. We got a question in the chat. What about indicator validity in times of crisis? Do they still work? Yeah. Yeah. Interesting question. I think what we see in many companies right now, probably also given the various situations we've been in, there is a lot of focus on the short term. So I see more and more companies differentiating between the SNP forecast and the short term forecast. And I think here, there are these different models. You can use different inputs. I believe you should be careful with re-evaluating the assumptions all the time for your SNP forecast. I think that should remain fairly stable. But instead, you should consider how do you model the short term forecast as well to consider that part. So I think instead of re-adjusting everything in the SNP forecast, there is more about is there another forecast you can use for the short term, taking into account, you could say, the current climate and the next six months horizon maybe. I don't know, Pierre, what your opinion here is on this. The various crises, whether that make you consider re-adjusting these indicators more frequently. Yeah. No, I don't think so. I also think when we did the selection of the macro indicators, it was also a long time period of data that we used. We used many years. I think I can't recall if it was 10 or 15 or whatever it was. So I think there is some kind of stability to the indicators that are in scope and not go and change them quarter by quarter, depending on what's happening in the outside world. So at least I think we should have trust that they also kind of last on the long run. Cool. Yeah. Cool. Yeah. I hope that answered that question. Maybe if we focus more on how to give you practical insights into how to get started, how to use this. We had a question in the chat asking, how can you actually incorporate those macroeconomic indicators without making your forecasting more of a black box and making it too complex? Yes. What's your perspective on that, Mike? Follow the steps provided by Pia, because I think I see this as a chance to get some outside in perspective and a way to challenge, you can say sales and the statistical forecast. So I think step one is to use it as an index and not directly embedded in your forecasting process or in your statistical modeling. You can say then step two could be considering other some markets where there is a strong correlation and causality and can then derive the trend directly from the indicators and apply on top of your baseline forecast. That could be considered. Then you can say where to start. I think it's important here that now we're talking about macroeconomic indicators. And as Pia also mentioned, the more years of sales history you have, the stronger foundation you build. So that also normally means that you need to start with markets where you are fairly mature and not in markets that are emerging or where you just entered. Because there normally you expect a growth. And then if you do the correlation analysis here, then even if the market is stable, there will be a growth. So that means that that growth will just extrapolate through the roof. So start out with markets and product categories where you are fairly mature. So if I understand it correctly, if you want to get started first, don't start by incorporating it in your statistical forecast. Do it simple so that you can understand what is the relationship and have the right conversations and start with a market where you're fairly confident about the business situation and then look into further topics. Yes. And then really, yeah, as you also said, Marie, what are the conversations you would like to have in your DM meeting? So you need to bring them forward there. Right. Cool. I would like to take one last question, which just came in the chat to you to Mike and Per. And that is how did you arrive at X number of number of indicators? The task of selecting five, ten or a hundred can be difficult. Where is the sweet spot? How did you to approach that in the in the project? So I don't know if you start here or if I should. I think we did this by modeling and also how well does it kind of explain, but also taking into account the complexity where our models like R squared. And I think Mike, you probably know more kind of advanced models to go through. But I think this is what the model balance, how much explanation does it do and also taking into account how many dimensions is it looping in and then balancing out. But that's the approach of each other. Yeah. And what we see here, both from your case, but also from other companies, ballpark between two and five, I would say that is probably where you have the biggest benefit, because you also need to be able to to explain it and understand it afterwards. So, yeah, as I say, that's the that is probably where you should start and rather have two where you can actually explain the causality than having ten where some of them is more vague. Yeah. OK, thank you for those insights. I think that goes back to the design principle. Start simple. But we have reached the end of our webinar for today. I hope you had a great session getting some new insights. If you have any burning questions, please still feel free to put them in the chat. Thank you. Thank you. Thank you for taking the time to and taking us along the Grundfos journey. Thank you to Mike for for hosting together with me. And I would like to just say goodbye to you with a little outlook that actually this was the first part of a webinar series on demand planning. So we would just like you to stay tuned for the next invites coming your way on further interesting topics on demand planning and how you can improve the quality of your demand planning even more. Thank you for joining today. Thank you. Bye bye. Bye bye.