This paper discusses harnessing omnichannel customer feedback data using deep learning, and integrated with NPS survey data to deliver superior customer experience (CX). While there is lot of work done and tool availability in this field for English, there is a lack of algorithms and research done for local languages, such as Bahasa Indonesian. We describe our Quasi Recurrent Neural Network (QRNN) based approach that we used on customer omnichannel feedback for topic modelling, priority prediction and sentiment classification. We further demonstrate the use of our work for multiple clients in Indonesia, in order to set up an advanced CX monitoring and improvement programme (CXM). We showcase our holistic approach to customer experience, integrating omnichannel customer feedback data with NLP-based predictions. We also showcase two cases where we deployed our intelligent CXM tool and set up intelligent conversation monitoring systems to predict customers at the verge of churn-out, beforehand.
Mobile devices are everywhere, and changing everything. Nowadays, people simply swipe or tap their screen to share their opinions with the world. Decisions are influenced by feelings, moods and our memory. As a research industry we have to adjust to this new reality if we want to stay in touch with who our customers really are. Measuring purely rational processes is no longer good enough. Instead, bridging rational and emotional drivers is key to accurately predicting how consumers behave and decide. Inspired by these developments, we are introducing Unspoken?. By combining intuitive mobile techniques with psychological theory and advanced modelling, this mobile-only implicit application opens the door to a new generation of research solutions.
This paper presents a combination of two research techniques to better understand consumers (scientific database), and to help marketers create better product and service concepts. The two research techniques are conjoint measurement for stimulus development and data collection, and segmentation and/or optimisation for database development and application.
This paper discusses a large research project conducted to investigate factors impacting on the relationship between banks and their business clients in Australia. A survey was conducted during February - March 1999 with 276 executives and financial managers on their firms relationship with banks. Factor and regression analyses were used to identify the underlying factors of close and long-lasting relationships between banks and their business customers. The insight gained through the data analysis guided the process of model building using a structural equation modelling techniques. The model shows that the measure of trust is crucial to understanding bank-business customer relationship.
The paper describes a proprietary modeling process which has been used to isolate advertising effects of a short-term nature and to document long-term effects. Many of the findings of this work are congruent with the findings of John Philip Jones, particularly the clear-cut ad effects in the week of exposure. However, this model makes use of standard household panel data rather than the "single source" data used in Jones analysis and unlike Jones, offers a concrete mechanism for examining long-term effects.
This paper presents two case histories showing how world-wide segmentation of consumers aids in new product development. The segmentation approach was originally used with sensory analysis of products, but has been expanded to concepts on a transnational basis. These segments transcend countries - viz., the same segment exists in different countries, but in varying proportions. The interview uses multimedia concepts, designed by experiment. The analysis uses modelling and segmentation to create individual concept response models. The result is a database which can be used to understand the segments and to create new concepts in a rapid fashion.
As international trade grows and consumer are faced with an increasing proliferation of foreign brands in their domestic markets, they increasingly use country of origin cues as an aid in evaluating products. At the same time, producers needing to establish unique positions in highly competitive markets, and government needing to protect domestic producers from imports, make increasing use of a variety of product origin cues in advertising, packaging, labelling, and branding. In spite of growing interest in the effects of the made-in concept on buyer behaviour, little is known about the correlates of product images and about the effects of such images on behaviour. This study proposes a model of the country of origin effects. The model was tested using LISREL, a second generation analytical procedure, on data drawn from a large-scale cross-national study of made-in images. The model suggests that buyer behaviour is affected by consumer evaluations about various foreign products, that product evaluations are dependent on consumer beliefs about these products' quality, and that these beliefs are strongly affected by consumer familiarity with products from a given country, and by beliefs about the country itself and its people. Strategic implications arising from the model, which was validated in eight separate tests, are discussed.
There has been growing interest in recent years in the tracking of advertising effects on sales and brand awareness through the use of econometric techniques. Typically, however, these approaches using Koyck transformations suffer from the problem of autocorrelation within the data. This paper presents an alternative approach, commencing with the removal of systematic variations in the dependent variable through the use of ARIMA modelling techniques. The combined approach is termed ADTRAC. This paper presents three examples of ADTRAC modelling for a major national UK retailer.
The purpose of this contribution to our seminar on modelling is to make two points: 1. There is usually little point in modelling something if we do not know what it is; 2. Developing the required empirically-based generalisations in marketing is both possible and essential. To illustrate, we summarise an experimental study on pricing carried out in the UK in 1986. Our aim was to establish whether the sales response to a given price change would generalise.
This paper concentrates on the Brand/Price Trade-Off (BPTO) modelling technique; how in the context of growing interest in price testing methods it evolved from earlier ad-hoc pricing work, and some of the problems that it helped to overcome. Since it was first used in the early 1970s, the method has become increasingly sophisticated, and adapted (largely by the author and his colleagues) to an expanding repertoire of market types, representing differing purchasing decisions by consumers. In addition to an account of the historical development of the approach, the paper also provides a description of the state of the art in terms of data collection and analysis. The paper continues by considering different types of purchasing decision in differing markets, and how the model is adapted for use in these. Finally, the paper concludes by reviewing the strengths and weaknesses of the method compared with others commonly used, and provides some suggestions for future development.
This paper takes as its thesis the argument that the dynamic changes taking place in agricultural markets and marketing offer agricultural market research a chance to enhance its worth and standing through the provision of advanced marketing aids developed in the world of fast moving consumer goods. It warns, however, of the dangers of adapting such techniques to agriculture without sufficient understanding of the special characteristics which affect each of our markets. In doing so, the paper takes a number of case studies involving decision-models, pricing models and product positioning techniques and investigates for each the special circumstances which need to be borne in mind for their adaptation whilst at the same time, highlighting their potential for widening our vision and understanding of the markets concerned.
The paper stresses the need for modelling as a tool for structuring the complexities of the marketing environment so that marketing management can better utilise the data at their disposal as well as their knowledge. This will help in meeting the goals of the organisation. The paper further looks at some available models and their implementation and use.