May 2008 | By José Almagro. Executive Chairman. Bayes Inference S.A. and Professor at IE Business School

Companies that adopt a modelling strategy to anticipate, identify and satisfy customersâ?? requirements will not only obtain basic competitive advantages over companies that do not, they will end up pushing them out of the market.

My theory goes like this: companies that adopt a modelling strategy to anticipate, identify and satisfy customersâ?? requirements will not only obtain basic competitive advantages over companies that do not, they will actually push them out of the market.

The modelling strategy is not just another strategy in a repertoire of feasible options, but rather a controlling strategy. This dominance is due to two coordinates. Firstly, given a certain set of activities, the use of models produces optimum combinations and is therefore the most effective way of organizing said set of activities.

Secondly, companies equipped with this strategy are more innovative and reduce the costs associated with misguided innovations. In short, operational and economic efficiency, early diagnosis of market, customer and policy trends, and innovative drive constitute the three key factors that turn a modelling strategy into a controlling strategy.

I maintain that companies, which are a relatively modern product of economic evolution, naturally possess behaviour adaptation mechanisms that give them a certain characteristic mould. Two kinds of occurrences make possible a new kind of corporate adaptation similar to the scientific adaptation of societies as a whole, which we refer to as the modelling strategy. These two occurrences are:
1) The information and communications boom, coupled with the exponential growth of processing and information storage capacities, and
2) The extraordinary development of statistics, modelling and decisions in areas that are dynamic and massified, with outputs that are defined both quantitatively as well as qualitatively.

If my theory is right, companies should roll out a purposeful modelling strategy as expeditiously as possible, as the competitive transformations we forecast will render the profundity, scope and quality of the modelling processes a critical factor in growth and survival.

Models, Decisions, Companies

Given that so far we have been speaking about science, you may think that this has little to do with the marketing problems faced by companies. Hence, I shall now list some of the problems that I have come across where a model provides the solution. A media publishing company has to distribute its product every day at thousands of points of sale. At each point of sale it has to decide on the number of copies to minimize the cost of excess product or non-sales if there are insufficient copies to meet demand.

A company in the telephone market has to deploy a network of huge capacity to provide Internet, TV, VoD, home automation and other products. The deployment and the order in which it is carried out are conditioned by demand, which must be calculated in small geographical units. A soft drinks company has to decide on the amount of advertising it places with the media, the distribution among the different kinds of media, timing and advert positioning.

A large fashion retail chain that handles hundreds of thousands of references must manage the price reductions at each one of these outlets to prevent loss of income through unnecessary price reductions or losses through depreciation of merchandise because of overcautious price reductions.

A company with several million subscribers has to decide on a loyalty action plan for subscribers that are likely, up to a certain point, to transfer over to the competition in the next three months. Applying loyalty measures to customers with little likelihood of moving away entails turnover losses. Failure to apply them to customers that will probably migrate entails loss of customers and also loss of turnover.

In all these cases we are faced with dynamic markets and with decisions, almost always large-scale, multifaceted on occasions, and usually repeated over time, where we have a more or less broad range of alternative options. Each option offers uncertain results; sometimes these results are quantitative, other times there are output variables that are quantitative. And each result has a cost, although this cost is not always known, i.e. it is also subject to uncertainty.

Actually, all decisions have an identical structure: A set of options and, conditional to each option, a structure of uncertain results, and a utility or cost function, in such a way that we can convert each consequence into a certain cost or utility. In each case there is a model function that we have referred to as the direct economic function or optimising function.

Models reduce the uncertainty of the decision taker with regard the results associated to each of the available options, and even provides a description of the probability function of the consequences. This description enables decisions based on maximum expected utility or minimum expected cost.

But what really is a model?

A model is a representation of reality which, applied to markets, customers and processes, enables the result of actions to be forecast and measured, making it possible to be aware of the potential impact of different alternatives and scenarios, to forecast actions of maximum utility and, therefore, to optimise the decision-taking process.

Models therefore express causal relations in a probable and dynamic way, and must use all existing information. Put another way, the models described here are predictive and explanatory and both features are closely tied.

There are at least four ways in which models contribute to diagnostic selection and assessment.

Firstly, models help with the early removal of ideas that belong to pre-modelling stages. Secondly, models enable policies, products and activities introduced in our interaction with the market to be measured. Thirdly, models permit market trends to be forecast, putting pressure on the need to innovate. And finally, models detect, through forecast errors, anomalies in the performance of the related markets, either through innovations of which we are unaware or through shortfalls in our own interpretations.

Companies equipped with this strategy are more innovative and reduce the costs associated with misguided innovations. In short, operational and economic efficiency, early detection of market, customer and the policy trends, coupled with a drive to innovate are the three distinguishing factors that make a modelling strategy into a controlling strategy, which leads to enhanced bottom line results driven by optimisation.

About the author