Simulations are an important element in business planning, enabling the effects of changes to be quickly assessed. Based on a planning model, input parameters such as quantity or price increases are varied to determine the consequences for specific output variables, such as sales, costs, or margins. Possible starting points for simulations in business planning can be the “Integrated Financial Planning for SAP S/4HANA” business content or customer-specific planning models.
SAP Analytics Cloud (SAC) includes specific tools for simulations, such as the value driver tree and the compass. In addition, simulation scenarios can also be created in SAC using standard tools such as stories and data actions. To compare the individual simulation methods, we implement a simple simulation model using the different methods and compare the results in a kind of decision matrix. This provides companies with guidance in selecting the appropriate simulation method for the problem at hand.
The simulation model calculates product sales for two products based on sales volume and sales price. A global price increase and the increase in volume per product serve as simulation parameters.
The value driver tree presents causes and effects in a clear graphical overview. The calculations are modeled using calculated and restricted key figures and visualized in a tree-like structure. The simulation parameters are basic key figures on which the calculated key figures rely. They are, in a sense, the leaves of the tree, and their values can be changed manually directly in the tree. The price increase affects both products, while the increase in volume per product can be entered separately. The simulation result (sales) and all intermediate results are updated immediately.
The scenario already reveals a limitation of the value driver tree: If too many objects (in our case, the products) are involved, the tree can quickly become too large and confusing. In addition, the simulation logic is limited because the entire simulation model must be implemented using calculated key figures.
The compass is based on the same simulation model as the value driver tree. This means that here, too, the simulation parameters are basic key figures and the simulation result is a calculated key figure. Unlike the value driver tree, however, not the entire tree is displayed, but only its leaves and roots. Instead, interval ranges can be defined for the simulation parameters so that the system automatically calculates the simulation result for a wide variety of parameter combinations (Monte Carlo simulation). Consequently, the simulation result is not a single number, but a range with a typical probability distribution.
The three simulation parameters with their intervals are listed in the left-hand section. The simulation result with the probability distribution is displayed on the right-hand side.
Instead of using calculated key figures, it is also possible to implement the simulation logic in a data action using an “advanced formula.” This variant has the advantage that it can also map more complex calculation logic. The simulation parameters are maintained in input tables, while the simulation results are displayed in a table or chart. For larger and, in particular, multidimensional models, the method is not subject to the restrictions of the value driver tree or the list in the compass, but offers all the filter and drill-down options of the pivot table.
The upper section contains the basic data and the tables ready for input for the simulation parameters. The calculation is triggered by a data action and the simulation result is displayed in the lower section. The simulation results match: The total sales (5,345.60 euros) in the value driver tree correspond exactly to the value calculated in the story by the data action and the mean value of the Monte Carlo simulation.
The following table compares the different simulation methods based on various parameters. It can be used as a decision matrix to determine which implementation is best suited for a specific simulation task.
For simple scenarios, either the value driver tree or the compass with their respective specific visualization options are recommended. For more complex cases, the conventional solution with stories and data actions is probably the better choice due to the large amount of multidimensional data.