There are four pillars in our modeling process: identification of the problem, data availability, modeling ability and technology for statistical analysis.
» Model requirements
-
Management problem
Our model must have its origins in correctly identified problems, so that they provide answers to clear questions, such as: what is going to happen next? What is going to be the evolution of the KPIs of the company? What is going to be the evolution of demand? What results, in terms of cannibalization, should I expect from introducing a new product? What are the expected results of a customer loyalty program? Our ability to answer these questions depend on our ability to quantify and measure strategic decisions and those taken by competitors, as well as other market innovations.
-
Detailed historical information
It is not possible to build a proper model without detailed historical information. The more information available, the greater the chances to build informative and useful models. In fact, the dynamic structure of information is essential in order to identify causes and effects and generate explanatory models. Furthermore, information is not uniform around the whole data set, but is highly concentrated around anomalies or outliers. The concept of an outlier in itself is a great source of information, since it allows us to distinguish between ordinary and anomalous behavior. It is also obvious that we must be able to establish a model for seasonal cycles, which is attained with the information derived from data on several of these periodical cycles.
-
Modelling ability
There is no doubt that the most scarce resource is modelling ability. We must notice that this ability increases with modelling experience and the previous knowledge of the business and sector. Model building is an important task that cannot be carried out by businesses themselves, but it must not be completely outsourced, given that business knowledge is essential.
Model building is an iterative process fuelled by data analysis and business knowledge, and fully automatic model building procedures are contrary to our philosophy.
-
Model building procedure
Market modeling is powered by a learning process based on data analysis and critical reasoning. More»
-
Dynamic Demand Attention Systems
Our solutions are customized for each of our customers, adapted to their needs and designed to optimise their decision-making through time. More»
