Sentiment Analysis & Customer Satisfaction
Customer satisfaction has been one of the most primary aspects of any business ventures. Resolving the needs and wants of the customer is possibly the most central element for the concept of a corporation or a business.
The importance of customer-company interaction is well-known, the most streamlined and satisfactory it is, the more growth and profit a company accumulates.
India
Client
Online Delivery Industry
CLIENT
Our client is a grocery delivery company with an ever-expanding customer base with support systems already present. We were requested to provide a new medium of interaction within their already existing application; this messaging service will be a bridge between delivery agents and the customer. The task is here to analyze the messages received and create contextual and simple replies to boost efficiency, coherence, and customer satisfaction.
CHALLENGE
There are certain challenges that need to be tackled for definitive functionality. They are listed below:
-
Language Proficiency - Most of the delivery agents aren’t usually familiar with the lingua franca of the user base, which is English. The replies sent my delivery agents might seem improper to the customer because of their simpler and broken English.
-
Location Verification - Most of the time, the biggest hurdle and point of annoyance for many customers is verifying the delivery location. An automated reply system can definitely help to identify directions with more ease.
-
Simple Communication - The presence of convoluted messages can anger the customer, simpler messages can reduce this problem and keep the conversation succinct.
SOLUTION
Solution summary:
- A quick reply system has to be created within their already existing application to boost communicative action between customer and the delivery agents.
-
This system needs to cover the needs of the customer and also be easy enough for the delivery agents to use.
-
This chat feature has to be self-contained in resolving the requests of the customer.
-
Simple language and interfacing is a must.
By ideating on the problem, we have resolved the major challenges and added features that will make the experience more suitable and well-rounded:
-
Conversational data is gathered and archived for analysis, pertaining to interactions between customers and agents. The feedback by prior customers are categorized and also compiled.
-
Three models are developed utilizing the inference arrived at after parsing the data.
-
A customer sentiment detection model is created first, which detected the moods and the context of the messages sent by the customer.
-
Main intent detection model is created for identifying the absolute wants and needs of the customer, pinpointing what has to be resolved.
-
By integrating the two previous model, a chat recommendation model is created. This model synthesizes replies by finding out the intent from sentiments and contexts.
RESULT
Implementing a model like this boosts customer relations and reduces redundancy. Improving customer relations from the lower level can promote a positive net gain for the company. Quick resolutions and agent training can also be boosted to the maximum with this solution. Contextual replies can also promote cordial relations between agents and the customer, promoting a positive orientation. With everything organized within a database, any request can be resolved with ease.