Sales and Operations Planning (S&OP)

Introduction

S&OP (sales and operations planning) has never been more important in the wake of digitization. Some companies are trying to catch up with digitalization and some want to be the pioneer. In both cases, our DSML (data science and machine learning) platform will bring them a significant step forward. It’s all about faster and easier use of data science.
The platform covers the whole analytical spectrum, from descriptive and diagnostic to predictive and prescriptive.

In this case study we will mainly deal with “What-if”, “What-next” and “How” scenarios. We will show you how you can optimize your contribution margin with our platform.

Initial situation

Let’s jump directly into the application and have a look at our initial scenario. A scenario in PlanNow is a selection of the data you currently work with. The initial scenario is called “Forecast 2018”.

The structure of a scenario is based on a model. To get an overview over the data we have in here, we can open the model editor. A model contains three important layers. The data passes through all three layers during scenario creation.

The first layer is called “Importer”. Here we can define the sources our data comes from. This can be Excel files, CSV files, Microsoft SQL databases, MySQL databases, RESTful APIs, and SAP systems.

The second layer is the “core” of a scenario, it’s called “Model”. Here, we describe how the data is mainly stored in a scenario.

The last layer is a sort of cache, it’s called “Datasources”. Here we can process the data for dashboards. It makes sense here to combine tables that belong together or to perform some analyses in R. This makes the work in dashboards much easier and gives us more possibilities.

Whenever we change scenarios, whether through simulation, transformation, or modification, we change the “Model” layer. The “Datasources” layer gets recreated.

You have now seen the data flow in a model through the important three layers. There are, of course, other things that can be defined in a model, which we’ll get to another time.

To look on some data our scenario contains, we can open a PQL query tab and run some queries. PQL is the abbreviation for PlanNow Query Language. It’s a language, for our custom in-memory database, which is used to query data.

With the following PQL statement, we can query the latest sales data from our scenario.

Based on the sales data, a sales forecast for 2018 got created. This data gets queried below.

Those queries are useful to get a quick idea how the data looks like, but for now we need a straightforward visual representation of the data.

The bar chart in the lower right corner is comparing sales data from 2017 and the forecast for 2018. The upper left widget shows us the forecasted amount for 2018 and the sold amount for 2017. Every widget, except this one, can be used to filter the data.

Contribution margin optimization

Due to the expected increase in demand in 2018, we know that we will not be able to meet all the demand in that year. Therefore, we have developed a contribution margin optimization, using a mathematical model.

We have implemented it into the PlanNow Analytics Suite, so everyone in the company can perform the optimization without advanced technical understanding. They can easily pass various parameters through input masks.

To perform the optimization, we need to select our base scenario (“Forecast 2018”) and the simulation method, “Optimization” (that’s the name of the mathematical model). We can give a name and a description to the scenario that arises from the optimization. This is useful for documenting different runs.

Let us look at the parameters we pass to the optimization. The first two parameters, Start Date and End Date, define the time interval in which the contribution margin will be optimized. “Factor Stock”, “Min Stock”, “Max Stock” and “Num. Periods Stock” say how different stock levels should be treated. “A customers”, “B customers” and “C customers” define the priority of the particular customer categories. “Ecological impact” is another factor that can be considered during optimization.

Let’s execute the optimization.

Contribution margin analysis

After the optimization is finished, our new scenario is created. By expanding the scenario, you can see the KPIs (Key-Performance-Indicators), which were defined in the model.

“Cost”, “Revenue”, “LineUsage” and “Drops” are based on the optimization. “Drops” specifies the sales quantity that should not be performed according to the optimization.

Let’s take a closer look at the result inside a dashboard.

The lower left widget shows you some KPIs. The widget above displays you the optimized contribution margin. The bar chart on the upper right visualizes the utilization of the production lines. We can see that all production lines are fully utilized. The lower right bar chart compares the sales from the previous year, the forecasted sales, and the optimized sales.

Production planning

Through the contribution margin optimization, we now know the demand we will cover. Using other simulation methods, we can successfully implement What-Next scenarios. We have developed a heuristic production planning and implemented it into our platform. Let’s execute it based on our contribution margin optimization.

Let’s look at the data from the new scenario inside a dashboard.

The upper right widget displays the production plan for the individual production lines. To get a more detailed view, you can scale down into hours.
The lower right chart allows us to gain some insights about the production plan. Simply by clicking on a campaign, you can see more detailed data.

We created, based on a forecast and on a contribution margin optimization, a production plan for the whole next year.

What-If scenario

Our powerful data science and machine learning (DSML) platform gives us the possibility to compare What-If scenarios simply by modifying an existing scenario.
Let’s see what happens when we increase the throughput by 10% of a production line, named “HH”. We remember, from previous data modeling, that we need to know the line ID to modify the production line. Therefore, we open the model editor again and query the table with the production lines.

We can see that “HH” has the ID 0. So, we can move on and transform the scenario.

A new scenario resulted from the transformation. Now we’ll run the contribution margin optimization with the same parameters as previously on the new scenario.

Let’s compare the KPIs from the first optimization and from the modified one. We can see that by increasing the throughput, we have about 346 tons less drops.

Let’s run the production planning and then compare both scenarios in a new dashboard.

The upper left widget shows how our contribution margin changed through increasing the line throughput. Our CM increased by 1.81%. The widget nearby shows that the productivity of the line increased by 9.7%. The widget below compares the different stock levels of the whole year. The upper right chart compares the drops, and the chart below compares the productivity of all four production lines. We can see the increase at HH, but also a small decrease at HS and HT.

Summary

This case study showed you how PlanNow Analytics Suite can be used to drive your business forward with easier use of data science. This revolutionary platform helps you to gain insights from your data and provides you with data driven decision-making. It can increase your contribution margin by 3 – 5% without additional capacity.

We have shown you a snippet of our model editor and hope it has given you an idea of the many possibilities you have in there. Our platform offers many collaboration opportunities that we hope to show you another time.

To perform everything, we have shown you, no great technical knowledge is required, so even business people can work with the tools we provide.

We hope to welcome you soon on our pioneer platform. If you are interested, just send us an inquiry. Feel free to check out our blog for more case studies. See you next time.