Agent Based Simulation Marketing Mix Model for Budget Management in Cosmetic Industry

The world is changing with the new technologies available, and so are the markets. Dealing with these changes is the most difficult challenge that every enterprise has nowadays. Companies invest more money every year in marketing efforts, without accurate measures of their effectiveness on the customers. But not only the touchpoints were renewed, also the marketplace. Online shopping is an effective and cheap way to deliver products and it is highly used in almost every product-based company.

The appearance of social networks as a brand-new marketing channel made the companies invest lots of money on these platforms only to be a part of this new world, not to be kept aside. Despite the social networks improved drastically the targeting of every campaign and started to provide hard data-driven techniques to add value to the marketing process, they are still channels that are very difficult to manage. Nowadays there is a huge variety of different social networks to put the marketing budget on, but also there are different ways of investing on them. Although social networks are important, as in an investment portfolio, companies need to diversify. There is still a wide variety of channels to get to customers and is vital to explode them all at their full potential.

AdSim is an agent-based Marketing Mix Model developed in the Anylogic platform and applied to Cosmetics Industry. Its main goal is to estimate the Return Over Investment (ROI) of the marketing budget of the company in study by predicting customers’ reaction to marketing efforts. The user selects for its brand the budget per touchpoint, the time frame of that budget and the amount of iterations the model will do to get statistically representative results. The model uses inputs from surveys made specifically to this purpose to parametrize the market. Every consumer on the survey is simulated in the model with its own parameters and characteristics. The results are affected at the end of the model by a scaling parameter to reflect the actual size of the market. As some of the processes within the model are stochastic, to ensure statistical representativeness, the model is ran many times with the same parameters. The output results in the distribution of the variables measured. To calculate ROI, investment effects must be isolated. For this purpose, a set of iterations is ran with no investments made by the user’s brand. The main assumption taken is that the competitors’ investments will remain the same.

Postes connexes