This research tested the relationships between social media sentiment, customer satisfaction and stock price performance. Furthermore, social media sentiment was broken down into global or aggregated social media sentiment, customer-oriented social media sentiment and financial-oriented social media sentiment. Custom dictionaries were created by the researcher to classify tweets as customer-oriented or financial-oriented. The relationships were also tested for potential differences between utility companies and non-utility companies.
The research methodology employed was quantitative in nature and has been applied in a causal design as part of a longitudinal approach to the study. Secondary data were downloaded for 79 American listed companies for the period of 1 January 2011 – 31 December 2012. The data comprised of end-of-day daily stock prices, customer satisfaction scores from the American customer satisfaction index (ACSI), and tweets from the Twitter network that mentioned the company name or brand. A total of approximately 20 million tweets were downloaded and transformed so that it could be analysed statistically. Due to the large number of tweets that had to be downloaded and analysed, the researcher developed a Twitter scraper and sentiment analysis tool to do this programmatically. Sentiment analysis was performed by using a combination of two well-known dictionaries. Panel data analysis was done on the transformed data.
A statistically significant relationship could be found for the ability of global social media sentiment to predict stock price performance. Furthermore, it was demonstrated that financial-oriented social media sentiment predicts stock price performance more accurately than customer-oriented social media sentiment or even global social media sentiment. Aligned with previous studies, no statistically significant relationship was found between social media sentiment and its ability to predict customer satisfaction. Social media sentiment can however be used to analyse customer satisfaction in real time. Lastly, customer satisfaction is not considered in financial models predicting stock price performance. This supports the debate around the efficient market hypothesis (EMH) and its ability to take all information into account when valuating stock prices.
The researcher’s unique contributions to academic research are the way she developed the custom dictionaries, the large amount of tweets that were downloaded and analysed as well as the types of tweets that were tested in this research.