SageCX (Voice of Customer) : 4-Wk Pilot Implementation

Tredence Inc

Automate the extraction of trends across a portfolio of brands, geography and loyalty levels of customers using unstructured data sources

Objective: Develop an end-to-end solution to identify the emerging themes and subthemes in Travel & Hospitality space based on various customer engagement touchpoints using Natural Language Processing (NLP)

Key Challenges Addressed:

  1. Manual processing of data from different channels such as Emails, Chats, Case Notes
  2. Lack of understanding of the root-cause of customer concerns
  3. Tagging was limited to associate’s knowledge
  4. Failure to capture the latest trends timely leading to delay in addressing customer complaints and sub-optimal customer experience

How do we address your challenges:

Our solution will enable you to understand the customer sentiment by:

  1. Developing a robust Text Analytics pipeline that handles unstructured data from multiple sources and automated extraction of key topics
  2. Analyzing Voice of Customer across the customer journey path (Ex: pre-booking, during-booking and post-booking)
  3. Ranking emerging themes & sub-themes without having them pre-defined with options to customize
  4. Enabling contextual insights to identify newer opportunities for growth using Self-Service reporting across the portfolio of brands, geography, and loyalty levels

Pilot outcome: An automated tool that will ingest data from selected sources, conduct data pre-processing and modeling, and provide contextual insights

Implementation Plan: The break-up of the implementation plan is as below:

Week 1 (Discovery) • Conducting Requirements gathering with key stakeholders, understand data sources, business processes and the challenges associated. • Performing exploratory data analysis, sanitize required data elements

Week 2 (Data Preprocessing) • Feature Extraction: Enriching the metadata information by extracting features such as Brand, Loyalty and geography from the text • Text pre-processing: Extraction of relevant content, data cleaning by removing stopwords and standardize the content

Week 3 (Data Modeling) • Model generation: Generating themes & subthemes from the text corpus • Standardization: Formalizing output structure and break it down into different identity and mapping tables

Week 4 (Insights Generation) • Visualization of the insights and scheduling the process for a set cadence. • Reviewing and publishing the key insights and observations.

This implementation uses the these native Azure components: Azure Data Factory, Azure Data Lake storage (raw & curated data), Azure SQL DB, Azure DB, Azure Analysis Services, Azure DF & Azure DevOps