Unfuzzying the Fuzzy Front End: a methodology for feature selection in a global carmaker company

Figure 1: General view of FSM

Type de publication:

Conference Paper


Gerpisa colloquium, Paris (2016)


Automotive sector, Fuzzy Front End (FFE), product innovation



Innovative products are typically associated with high market and technological uncertainties and complexity (Holahan, Sullivan, and Markham 2014). In such cases, for which traditional managerial models tend to be poorly supportive, previous efforts oriented to reduce uncertainties may be performed before initiating the formal development process (Moenaert et al. 1995). In this context, the Fuzzy Front End (FFE) is presented as the earliest phase of the NPD process and include product strategy formulation and communication, opportunity identification and assessment, idea generation, product definition, project planning and evaluation, and executive reviews (Khurana and Rosenthal 1998).

Despite the progress that has been made in theoretical and empirical studies, the challenges to manage the front-end are numerous. Also, practical methodologies and tools to deal with such challenges tend to highly depend on specificities of the organization and its environment (Kim and Wilemon 2002, Oliveira et al. 2011). Notwithstanding, research in this area has been scarce and many authors defend that organizations need to develop proper FFE process and practices, which could fit to specific situations of the industry (Kim and Wilemon 2002, Oliveira et al. 2011).

The automotive sector is one of the most prominent when considering the adoption of formalized methods for New Product Development. In last years, the sector has seen the rising of a new approach: instead of conceiving very complex products to fit to several multidimensional customer demands, the design and implementation of new individual features becomes a very attractive strategy to allow customers perceive benefit and differentiation (Maniak et al. 2014). A feature is defined as an “identifiable aspect of the total offering that a critical reference group perceives and evaluates as an ‘extra’ to a known standard among comparable products” (Thölke, Hultinka, and Robbenb 2001).

This research is centred in new challenges that have emerged for the practice of the FFE in automotive industry due to the feature development perspective, which carries relevant questions to be answered in this phase: e.g. what features should be developed and, from those features available in current portfolio, which should be selected for a new product in particular. This study focuses on this second problem, from which the following research question was formulated: “Given a feature portfolio of a global automotive group, how to decide which ones should be included in a new model specification?”

As a result of the present work, a 4-stage methodology was proposed to face the feature selection challenge in a specific company. Such methodology seeks to integrate several information from both internal and external organizational environments, and intends to guide the efforts and decisions made in feature selection under a more objective and accurate approach. Finally, implications for theory and practice of FFE in automotive sector are discussed and new opportunities for research are proposed.


This study was performed in a 7-month action-research (AR) program, driven by key action research guidelines (Eden and Huxham 1996). The AR program was held between a large multinational automotive manufacturer and a research group of Federal University of Minas Gerais (Brazil). The main input data for the work was the global portfolio of features, which includes features currently applied in car models around the world and others in development stage. Due to the strategic importance and confidentiality of the portfolio, the company's research team previously prepared a codified, synthetic and static portfolio, based on the real one, which was used during the most part of the work.
The study followed 7 macro-steps: (i) Definition of research team; (ii) Adaptation of global feature portfolio; (iii) Mapping of managerial uncertainties and data-sources; (iv) Data collection and processing; (v) Design of the Methodology for feature selection; (vi) Simulation and tests; and vii) Documentation and training. In parallel to the steps iii to vi a computational tool was developed to support data processing and to automatize some analyzes.


The AR program resulted in a 4-stage methodology - the FSM (Feature Selection Methodology- Figure 1). This methodology seeks to integrate several information from both internal and external environments, and the output is a set of features to be included in a new vehicle model specification. The FSM turns the challenge of feature selection into a logical process of data gathering and analysis that reduces the inherent uncertainties and the subjective character of the decision-making.
Starting from the definition of a particular product model to which a specification template must be defined, the first stage consists on a market diagnosis of overall customer’s satisfaction in different feature’s functions. Then, the firm’s relative position compared to competitors is identified. The second stage adds product strategy into the analysis to identify performance gaps (what the current product is vs. what it should be). The third stage confronts this data with the number of features in current portfolio for each function. The main outcome of these first three stages is a set of functions that reveal to be more important for the performance of the concerned product.
The fourth stage aims ordering features from screened function according to some criteria such as similarity with current local applications, feature cost (per unit) and development time/cost. The ordered list, its subjacent criteria and the visual maps generated along the previous analysis constitute, in an integrated form, the formatted instrument to guide the managerial decision on what features to be included in product specifications.
This paper provides a detailed description of the proposed methodology, highlights how it fits to the case and discuss implications for theory and practice of FFE, especially in its approach to the context of the automotive sector.


  GIS Gerpisa / gerpisa.org
  4 Avenue des Sciences, 91190 Gif-sur-Yvette

Copyright© Gerpisa
Concéption Tommaso Pardi
Administration Juan Sebastian Carbonell, Lorenza MonacoGéry Deffontaines

Créé avec l'aide de Drupal, un système de gestion de contenu "opensource"