Social Analytics: Dimensions of Good Service

Thanks to social analytics, it is now possible to monitor, analyze, and interpret the interactions and relationships between people, topics, and content on social networks. By leveraging from social analytics solutions, companies gather and analyze users’ activities on forums and social media. Furthermore, social analytics can also include the monitoring of topics in order to see what is trending and decide how companies should act or react. A further application would be the analysis of social networks in order to identify significant influencers. Besides providing social analytics as a solution to the client, Arvato CRM Solutions have developped a consultative selling approach called Social Quick Scan.

But how can social analytics be implemented so as to optimize customer service? And which service qualities are most relevant? A research project by Arvato CRM Solutions and the University of Münster is on hand to answer these questions. The data for the analysis were gathered with the help of a survey of 647 German-speakers who frequently use social media.

Five Dimensions of Service Quality

Service Quality (SERVQUAL) has been widely used in both academic and praxis context to measure the service performance and the resulting levels of customer satisfaction of a brand. In the research project, following dimensions have been taken into consideration:

• Assurance: the extent to which employees are courteous and knowledgeable, i.e. their ability to inspire trust and confidence in their customers.
• Reliability: employees’ ability to precisely and reliably perform services.
• Responsiveness: employees’ readiness to provide help and fast service.
• Empathy: employees’ ability to focus on and care for individual customers and their needs.

The fifth SERVQUAL Dimension Tangibles (employees’ appearance, i.e. the state of their body and clothing) was discounted in the research project, as these tangibles tend not to be a factor on a social media platform like Facebook.

During the survey, the participants were asked to place themselves in a fictitious situation – an overcrowded and delayed train causing them to miss their connection – and then to seek help from the train company via Facebook. In response, each participant received replies from customer service with varying levels of service quality. For example, responses arrived anywhere between right away and two hours later (high and low levels of responsiveness, respectively). The participants would then give feedback with another post. Afterwards they rated their overall impression of the quality of customer service based on the individual servqual dimensions. Finally, the emotional states of the customers were ascertained using the so-called Sentiment Analysis.

An interesting finding of the research project was that the relative significance of the servqual dimensions in a social media context is different than in a traditional offline context: Earlier studies suggest that reliability is the most significant factor in relation to service quality. In the context of this study, assurance was found to be equally important. The third most significant was empathy, traditionally viewed as the least important dimension in an offline context. Surprisingly, responsiveness was found by this study to be the least significant, although response time is generally a significant metric in a social media context. One explanation might be that the participants exhibited different criteria for what constitutes a good response time. Some regarded two hours as prompt, whereas others felt that fifteen minutes was too slow.

Strengths and Weaknesses of Social Analytics Solutions

The research project also showed that social analytics, on a purely technical basis, doesn’t always provide satisfactory results. An analysis focusing on key words, for example, completely fails to recognize irony and sarcasm, often rating biting criticism (e.g. “a truly fantastic service…”) as a glowing review on the basis of the keywords being used. Solutions utilizing machine learning and artificial intelligence perform better in this case, although they tend to struggle with emojis and require a minimum length of text in order to function properly. And, lastly, the context of a post must be taken into account. When one user posts a negative review and a second agrees with the rating, the second post is still a negative review, despite using positive vocabulary.

For this reason, Arvato CRM Solutions employs, among other things, a comprehensive training of the Social Analytics Solutions, extensive test data and regular examinations by communication experts (including analysis of the results). Such a “hybrid” operating model ensures higher quality analyses.

Conclusion: Assurance and Reliability Make for Satisfied Customers

The study showed that assurance and reliability are the most important dimensions of service quality in a social media context. The perception of fast or slow responsiveness, by contrast, seems largely dependent on one’s own subjective definition and therefore can only be considered on a case-by-case basis.

It also seems clear that social analytics systems can only operate at full effectiveness upon multiple pre-requesits: a) if the algorithm is appropriately trained and optimized, both before and during the task; b) human involvement is necessary, especially for fine-tuning the results generated by the tools; c) regular monitoring of the performance to improve the quality constantly. It seems that tools cannot replace traditional quality control at the current stage; however, they can definitively expedite the process while making large quantities of data easier to manage – for excellent customer service in the online world.

Looking into the future, a hybrid operating model is essential for the social analytics solution – combining the expertise of the social media specialist and the technological advancement of the tools.

Author: Editorial team Future. Customer.
Image: © peshkova – AdobeStock