Leveraging AI to react to customer feedback automatically and manually
I am fortunate to share the news that my conceptual research on “Reshaping the contexts of online customer engagement behavior via artificial intelligence: A conceptual framework” has been accepted by the Journal of Business Research (Impact Factor 4.9). In the paper we develop a framework on how to start leveraging artificial intelligence to react to customer feedback automatically and manually.
We consider both firm-solicited (e.g. customer surveys and robocalls) and firm- unsolicited (e.g. reviews on social media and online brand communities) customer behavior as a stimulus for companies to serve their customers better.
In our framework we introduce the idea of linking the new information with other information sources (e.g. CRM databases) and processing the information in real- time with the help of AI to trigger both automated responses and help drafting manual responses.
Automated responses may stem from the known linkage between certain kind of feedback (topic or sentiment) and its linkage to known behavior of the customer pool. For instance, if the feedback from a Persona 1 falls into a category of Complainers, the machine may suggest providing the next delivery free of charge. For Persona 2, it may be enough to give 20% OFF coupon. It all depends on the information that the machine has about the customer feedback and the related behavior of similar customers. As the machine suggests certain responses, it learns from the customer responses and learns to give even better suggestions.
In addition to the thematic categorization of feedback, it may be important to understand the sentiments in the customer feedback. This requires statistical comparison of emotional charge in each message. As a result, the machine can learn that a certain level of emotions requires this kind of response.
In the paper, we emphasize that the response does not need to be automatic. It can be that certain feedback encourages human reaction to customer feedback. Here, AI can be anyway helpful in guiding humans how to respond to the feedback in order to boost sales or avoid customer churn.
While we started writing the paper with my colleagues before SecurAI was co-founded, I am happy to say that SecurAI is able to provide a practical solution to the concept that we described. For actionable results we need accurate topical analysis and a scalable multi-dimensional sentiment analysis. This is what SecurAI provides to its clientele as a routine service. Now we’re in the stage that we can start linking this information to CRM data, link feedback from different customer personas, and eventually start teaching machines on what kind of a response is best for each solicited or unsolicited feedback. I am also happy that my research nurtures SecurAI’s development, giving it this market-relevance and expert-solution edge.
I’m looking forward to writing the next paper on the topic, and sharing information how the conceptual framework helps develop real-world solutions.
The article was co-authored with my colleagues at Henley Business School (Rodrigo Perez-Vega, Cristiana Lages and Niloofar Borghei Razavi) and Feedbackly (Jaakko Männistö).