Problem:
In 2008 there were three huge companies in the telecommunications market in Argentina. Each company sold some of the four products (known as “quadruple play” when combined), including high speed Internet, cable TV, telephone, and cell phone services. Many households purchased individual services from the different companies. The aggressiveness of the players made “product bundling” a key marketing strategy, but there was not a clear understanding of how to use this strategy, or the impact it might have on the market.
Each company wanted to enter the markets that they hadn’t covered yet, but each one knew that penetrating the competitor’s niche would force them to reply in the same manner. The consequences of such actions had to be estimated. That’s why Telefónica (one of the competitors) decided to employ Continente Siete, a consulting company, to build a model of the market using AnyLogic multimethod simulation software.
The challenge was to build a model that would allow the client to analyze several scenarios, taking into account the whole telecom market (with its 3 major players) and the “product bundling” effect. Scenarios had to cover the telecom market evolution for 24 months and issue an integrated set of metrics (market share for each product, revenue, brand effect, etc.). The model had to consider all 100+ products potentially available in the market. It also had to work with a previous conjoint analysis made specifically for this purpose, as well as include historical information on complementary products and macroeconomic variables.
Solution:
The core of the model was consumer choice. Households were modeled as agents (1 agent simulates 1000 people). The whole consumer decision process (Know-Evaluate-Decide-Implement) was embedded into each consumer agent, based on the results of the conjoint analysis. Many factors influenced consumers in their decisions. For example, the conjoint analysis allowed the estimation of each product’s attribute value (including both price and brand) for each consumer. Besides those preferences, the consumer had to be familiar with each new product and keep it in mind before buying. Also, wallet restrictions were added, as all of the agents were divided into three groups by their income level. Other barriers, such as needing to buy a PC to have Internet, going through the hassle of changing the service provider, and even the service provider’s retention strategies (making a better offer to a customer who wants to quit the service), were also taken into account.
The companies were modeled as agents too. They dealt with different levels of the four basic products (for example, low-speed Internet vs. high-speed, etc.), bundles, promotions, and retention offers. Prices, promotion lengths, retention policies, and even the launching of new products were all controlled by the companies.

The whole model reflected the current state of the market, including the companies, their income, their number of customers (structured by geographical and socio-economic properties), technologies sold by the companies, government regulations implied, and PC market evolution over time. All macroeconomic variables were simulated with the System Dynamics approach. An adapted innovation diffusion Bass model was used to model PC evolution over time.
Scenarios were built to understand possible outcomes of launching the new basic products for each company. A particular emphasis was placed on possible delays between the competitors’ product releases. Price escalation and promotion removal scenarios were also tested. Another key question was the subsidization of buying PCs for low income consumers. Scenarios were built to understand the impact of this policy.
Outcome
It was the high flexibility of AnyLogic simulation software that allowed the consultants to grasp the system in its complexity. It allowed them to build a truly multimethod model using System Dynamics for reflecting the market as a system and Agent Based Modeling for simulation of the behavior of each household and of the companies.
The use of the validated model for scenario generation allowed Telefónica to build up their strategic plan for 2009 to optimize their performance in the market.
Watch the video of J. Pablo Rodríguez Varela, the co-founder of Continente Siete , presenting this case study at The AnyLogic Conference, or download his presentation.
