Problem
One of the world’s leading motor vehicle producers needed a strategic forecast of their performance in the US market for the next five years. The company wanted to estimate the dynamics of demand on their product (motor vehicles of a particular class) and the expected revenue, taking into account current clients, dealers, competitors, and the used vehicle market. Their main objective was to determine how much product the company would need to produce in the following years. They employed The AnyLogic Company to create a complete model of the US market.
Solution
In the AnyLogic model, the whole country was separated into several regions, which were then separated into districts. All the geographical entities were considered independently. The districts were comprised of the people living in the area, the vehicles that some of these people owned, and the dealers that sold different brands.
The characteristics of the vehicles included brand (a total of nine popular brands were considered), market classification, engine capacity, model, price, and year. The vehicles for sale could also be either new or used. The AnyLogic Company consultants used the data on geographical distribution of used vehicles and their average price in each region, so prices for used entities varied geographically, while the prices for new ones were fixed all over the country. Dealers were considered to sell only new vehicles of a single brand. Each dealer operated within the region, and requested vehicles were always available for purchase.
Customer characteristics included all the essential parameters for a scrupulous marketing analysis. Customers were divided into segments based on their age, gender, and race. Characteristics such as state of employment and income were also included. The customer population was simulated dynamically. Deaths and births, together with other changes, were taken into account. Due to the large scale of the model, 1 agent simulated 100 people. All the input data on customers was taken from real life, including the official US unemployment forecast for the next five years.
Another customer characteristic that was considered was their attitude towards products. The whole model was based around this concept. A prospective customer’s desire to buy a product was influenced by advertisements, contacts with product owners, and visits to dealers (see the picture).
One of the main challenges was “translating” the data to the language of agent behavior. This required a detailed analysis of the facts and determination of the actual points that influenced customers’ decisions, their sensitivity to different factors, and the inertness of their decisions, etc. Thorough validation of the model based on historical facts and statistics was also important in this project.
Outcome
The client used the model created by The AnyLogic Company consulting team to draw up their strategic plan for the next five years and successfully align their marketing policies and production plans with the forecast.
Watch Anatoly Zherebtsov from The AnyLogic Company presenting this project at the AnyLogic Conference 2012 or download his presentation.
