Case Summary
The client, a New York based retirement solutions company, wanted to take investment decisions based on employee sentiments. They wanted to perform sentiment analysis on Glassdoor reviews but couldn’t do it manually due to intense volumes of feedback to be reviewed. SLK helped the organization extract insights into its culture and other defined themes to enable multiple business and investment insights using NLP and AI.
The Challenge
Glassdoor reviews, which are anonymous employee testimonials, are a great resource for learning about a company’s internal processes and environment. Company culture has been shown to influence overall company performance in the past, making it a useful early warning system. However, the client performed the sentiment analysis on the data manually which made the process slow and inadequate.
The customer needed a solution that could quickly extract data and insights from the Glassdoor data without human dependencies. They also wanted to have visibility on the data and insights for various teams and business needs.
The Solution
SLK suggested using AI/NLP (Artificial Intelligence/Natural Language Processing) to analyze Glassdoor reviews for a subset of companies available. NLP algorithms were applied to extract and quantify relevant information from these reviews. It also generated and aggregated scores at a company level for defined themes.
We extended the developed AI/NLP model to the full dataset and provided a basic UI for data visualization and analytics. We trained the AI engine to assist HITL (human in the loop) in validating against various defined themes.
Business Impact
SLK’s expertise in AI and NLP helped the company develop a fast, proactive, and scalable solution for sentiment analysis.
- We provided insights into the company’s culture, which can be used to create investment signals
- Proprietary ESG-based investment signal that delivers actionable insights for all investment teams
- Provided a thought leadership paper highlighting the approach to ESG (authentic and proprietary signals aligned with the client’s core values)
