Customer Feedback Classification and Analysis using Natural Language Processing (NLP) and Machine Learning
Analysing the customer feedback data and classifying them into a category
The majority of telecom service providers receive millions of customer requests daily, which is higher than ever before. It is becoming increasingly difficult to provide timely and efficient assistance due to:
- the high number of requests,
- the inability to go to physical stores, and
- the large number of employees working from home.
An omnichannel approach to communication, personalization, and immediacy is especially important during these times of crisis. A company that ignores these needs may face lengthy waiting times, annoying back-and-forth conversations with multiple executives, and inefficient automated responses. A customer’s bad experience with your company can ruin their rapport with you, and having an agitated customer is definitely not something you want. The number of subscribers and the variety of products and customized solutions have made operational tasks increasingly complex, since face-to-face assistance is no longer available.
A large amount of customer feedback data must be analyzed to classify them into categories, which is a tedious job. AI-powered tools can automate many tasks, but the real challenge is when a classification algorithm generates incorrect classifications. This can lead to a double task where the user will need to manually review and change the predicted category.
Therefore, machine learning models should be able to automatically correct themselves based on the learning process and update their algorithms to adapt to manual changes.
A change by Excelledia
Taking in to account the challenge as mentioned above, Excelledia brings forward the innovative change by introducing a smart feedback analysis platform for Telecom providers. The said feedback analysis is performed using the survey data collected form customers. This data is then analysed and classified into categories.
We also take pride in introducing our interactive dashboards, that provide analytics report on customer survey responses, TNPS rating, and keyword-based response categorization. It also provides information on the Machine Learning (ML) based response category prediction model and category recommendation model. User can also review, edit and update the customer response categories as and when required, and the algorithm makes auto corrections from the learning
ML-based solutions are able to classify with accuracy, and they also benefit from the fact that they can learn to identify new categories directly from the data itself rather than by manual coding.
Scope of this innovative change.
The feedback analysis platform contains the following list of features.
- The total responses KPI, which gives the total number of feedbacks from each customer (excluding the TNPS rating).
- Survey period that provides details on the period the survey was conducted.
- Current Month Responses, which provides current month response count and shows the percentage difference compared to previous month.
- Percentage count of customer categories such as promoter, passive, detractor.
- Drill down customer category into customer feedbacks and customer feedback word clouds.
- Filter responses by category, year month and day, and also view or edit matched category with responses.
- Category prediction using machine learning methods.
- New category recommendations.
- Review, edit and update the predicted response categories and recommendations.
- ML algorithm that is capable of making auto corrections from the learning.
- Word cloud that is generated to get the frequency of keywords.
- Set up AI model update schedule, and manual updating of AI model.