Integration of Voice of Customer into Customer Experience Measurement

The dissertation focuses on consolidated Customer Experience Measurement based on the integration of information mined from customer data in a textual form known as Voice of Customer (VoC). The main objective of this thesis is the creation of an artefact in the form of the Customer Experience multidimensional data model combining the information from textual VoC with other structured customer data extracted from various sources. The artefact enhances the currently fragmented methods in Customer Experience Measurement by providing more sophisticated data-driven and methodological approaches. In particular, text analytics is not yet fully established in marketing. The data model is divided into the textual and analytical part. Text analytics methods are performed during ETL processes. Following the Business Intelligence principles, different metrics are designed to track overall Customer Experience from the customer's perspective at each stage of the customer journey.Based on literature research, the author conceptualises the Customer Experience construct and defines its constituent elements. Customer Experience Measurement is further enhanced with detection of sentiment, emotions and personality traits – perceptual elements hidden in customers’ written expressions which emerge in their behaviour and accompany the entire customer journey. The author defines requirements on mining these elements with an emphasis on maintaining a customer perspective.The qualitative research among Czech organisations operating in the internet environment shows the immaturity in Customer Experience Measurement and lack of approaches to mine the information from textual VoC. The artefact is validated by Technical Action Research (TAR) in an e-commerce environment with real-world datasets. The author analysed textual customer data in the form of customer reviews with a set of text analytics methods – a lexicon-based approach for customer sentiment, rule-based approach and Latent Dirichlet Allocation for aspects detection, deep-learning for emotions mining, and clustering for personality detection. The results of methods met the stated criteria and were integrated with other structured data within Customer Experience data model. The proposed reports, which maintain the customer perspective, have successfully met the stakeholders’ goals. Beside TAR, seven expert opinions validated the usability of the artefact and assessed its managerial impacts. Experts agreed that the artefact mitigates the barriers of achieving the full potential of analysing VoC within Customer Experience mentioned in Czech companies. The artefact is thus properly designed and ready to be utilised by business practitioners.