19
Chapter 04 – Exploratory and Observational Research Designs and Data Collection Approaches
accuracy of the information is unknown.
• The sample of people interacting about the brand, product, or advertising campaign is a
self-selected sample that may not be representative of consumer reactions in the target
market. In fact, different social media monitoring tools often produce different results.
• Some social media sites are not publicly available for researchers to mine.
A listening platform or post is an integrated approach to monitoring and analyzing media
sources to provide insights that support marketing decision making (PPT slide 4-26). In the
past, larger companies often paid for a service that would read and clip articles from
newspapers and magazines. Listening platforms are a technologically enhanced version of this
older service. Reasons for deploying a listening platform are given below.
• Monitoring online brand image
• Complaint handling
Listening platforms are in their infancy and are ripe for a large number of research innovations
in coming years. The qualitative data available from social media monitoring can be analyzed
qualitatively, quantitatively, or both. Currently, most social media monitoring tools seek to
seamlessly mix qualitative and quantitative analyses. The earliest application of quantitative
methods is simple counts of mentions of keywords. Another emerging, but controversial,
quantitative tool is sentiment analysis, also called opinion mining (PPT slide 4-26).
Sentiment analysis relies on the emerging field of natural language processing (NLP) that
enables automatic categorization of online comments into positive or negative categories.
Initial research applied sentiment analysis tools to product, movie, and restaurant reviews.
Quantitative measures of sentiment are still limited as a large amount of data is currently
unclassifiable or incorrectly classified with current automation tools. But more advanced
sentiment analysis tools are being developed to go beyond grouping by category and enable
classification by emotions such as sad, happy, or angry. Thus, in the next few years, sentiment
analysis methods are likely to be improved substantially, with their use becoming more
pervasive.