Back to search

Athens University Economics and Business (AUEB)

Brand equity assessment: a computational model for mining consumer perceptions in social media

Abstract

dc:description

The proliferation of Big Data & Analytics in recent years has compelled marketing practitioners and business decision-makers to search for new methods for generating insights when faced to measure brand performance during campaign appraisals. Marketers are constantly asked by upper management to justify the impact of online marketing activities in terms of brand performance in the marketing mix. Financial measures such as sales and profit provide partial indicators of brand performance, leading marketers to also turn towards assessment of intangible market based assets such as brand equity. To address this gap, marketers over the past 20 years have introduced various approaches that assess key conceptual dimensions of brand equity from a consumer perspective. Prevailing methods focus on the multidimensionality of the construct using confirmatory factor analysis with structural equations modeling for evaluating their models. Although these measurement approaches have and continue to be used for assessment of brand equity during marketing campaign cycles, a well-observed challenge is that these methods heavily rely on traditional data collection and analysis methods, such as questionnaires, face to face or telephone interviews, and have a significant time lag. Recent years have witnessed a fundamental shift on how consumers choose to convey their experience during their interaction with a brand. The rise of social media platforms has empowered consumers worldwide to influence and shape perceptions of brands, leaving no choice to organizations than to adapt, invest in and monitor these channels. With mobile growth driving the change towards a connected consumer who with the power of a smartphone now has access to information previously only available through conventional means, marketing and technology are more than ever drawing towards a merge. Information is now digitized and stored—in the form of news, blogs, forum posts and social networks— making it increasingly important not to neglect such means. New computational tools that help organize, search, and understand this vast amount of information should now be used during decision making.In line with this paradigm shift, this study introduces a novel approach of CBBE assessment, through the application of big data and machine learning techniques in data collected from social media. We develop and introduce a conceptual model of steps that lays out a proposed method of how fundamental dimensions of CBBE, such as brand awareness and brand meaning, could be optioned from consumer perceptions, starting from a marketing perspective and describe the technical steps necessary to construct the measure. At the core of the proposed model lie topic and sentiment detection approaches to elicit the influential subjects that govern the generation process of CBBE. Instead of estimating brand equity based on low-level online social verbatim features (e.g. frequency of words with negative or positive sentiment polarity) we propose a high-level abstraction computational model which estimates CBBE based on the central discussion topics on online social networks. The model utilizes a novel genetic algorithm to address the problem of topic clustering in text data. The proposed topic clustering method is anchored on the Latent Dirichlet Allocation (LDA) probabilistic topic modeling framework, aiming at identifying cluster formations that are optimal in terms of semantic coherence. This work focuses on reformulating the clustering problem as a discrete optimization problem within the n-dimensional standard simplex since all the LDA-based data patterns correspond to n-valued probability distribution vectors. In this way, the NP - hard cluster assignment problem reduces to the problem of locating the optimal cluster-centroid positions within the n-dimensional standard simplex given that the number of clusters to be identified is equal to the number of topics to be extracted by the LDA probabilistic modeling technique. The novelty of the proposed genetic algorithm approach lies primarily upon the adaptation of the centroid-based encoding scheme, in the sense that cluster assignments are implicitly extracted by assigning each data point to the nearest cluster center.To illustrate the validity of our model we employ two case studies, in different business verticals, that collect brand related data from social media channels, such as Twitter, unveiling key CBBE dimensions during marketing performance assessments. In particular the first case study evaluates the model in the mobile telecommunications business sector, focusing on the two leading carriers in the USA, AT&T and Verizon while the second case study applies the model on the mobile app market, unveiling constitutional elements of CBBE relating to the UBER transportation network.The theoretical and managerial contributions of this research outline the need for marketing researchers and practitioners to adopt state of the art machine learning techniques and methods, when faced to assess specific marketing constructs, such as customer based brand equity. Our research contributes towards this end by outlining a step by step computational model that starts from a marketing perspective, describes the necessary technological steps and concludes with the insights that can be generated.

Degree

thesis:*
Grantor dc:publisher
Athens University Economics and Business (AUEB)
Year dc:date
2017

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Pournarakis, Demitrios
  • Πουρναράκης, Δημήτριος

Subjects

dc:subject × 11

Rights

Language dc:language
eng

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/40929

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Pournarakis, Demitrios; Πουρναράκης, Δημήτριος. Brand equity assessment: a computational model for mining consumer perceptions in social media. Athens University Economics and Business (AUEB), 2017. http://hdl.handle.net/10442/hedi/40929