Case Study - Sentiment Safeguard: Taming Tongues in the Rainbow Nation

Leveraging NLP to decode customer sentiment across South Africa's linguistic landscape, boosting Hollard's customer retention and acquisition strategies.

Client
Hollard
Year
Service
Data Science: Multilingual NLP Sentiment Analysis
Hollard sentiment-tracking dashboard with topic share, sentiment share and per-topic sentiment charts over time
  • Power BI

Impact

The implementation of this NLP-driven sentiment analysis system had a profound impact on Hollard’s customer relationship management and competitive strategy. By providing real-time insights into customer sentiment across South Africa’s diverse linguistic landscape, the solution enabled Hollard to respond promptly to customer concerns, potentially averting customer attrition.

The live PowerBI dashboard offered a visual, easy-to-understand representation of customer sentiment trends, allowing management to make data-driven decisions quickly. This real-time view of customer satisfaction levels across different product lines, regions, and languages provided invaluable insights for strategic planning and resource allocation.

The automated alert system for negative sentiment was particularly impactful. By flagging issues as they arose, Hollard could address customer concerns proactively, often turning potentially negative experiences into positive ones. This rapid response capability likely contributed to improved customer retention rates and enhanced brand reputation.

The competitor tracking feature added another layer of strategic value. By monitoring sentiment around competitor brands, Hollard gained insights into market trends, competitor weaknesses, and potential areas for differentiation. This intelligence informed product development, marketing strategies, and customer acquisition efforts.

Perhaps most significantly, the system’s ability to process and analyse sentiment across South Africa’s many vernacular languages set the system apart. This capability allowed Hollard to truly understand and respond to the needs of its diverse customer base, potentially opening up new market segments and improving customer loyalty among previously underserved linguistic groups.

Introduction to the Customer

Hollard is a prominent player in the South African insurance industry, known for its innovative approaches to customer service and product offerings. As a company operating in one of the world’s most linguistically diverse countries, Hollard faces unique challenges in understanding and responding to customer needs across multiple languages and cultures. The company’s commitment to using new technology to enhance customer experiences aligns with South Africa’s reputation as a leader in financial services innovation on the African continent.

Challenge

The primary challenge was to develop a system capable of accurately analyzing customer sentiment across South Africa’s numerous vernacular languages. This required not only advanced NLP capabilities but also a deep understanding of linguistic nuances, colloquialisms, and cultural contexts specific to each language group. Additionally, the system needed to process and analyze large volumes of data in real-time, presenting insights in an accessible format for decision-makers. Balancing accuracy, speed, and user-friendliness while accommodating the linguistic diversity of South Africa posed a significant technical and analytical challenge.

From the blog

Writing that goes deeper on the engineering behind this case study.

Choosing a BI Stack for a South African Business: Metabase vs Superset vs Grafana vs Looker Studio (and Power BI)

A practical comparison of Metabase, Superset, Grafana, Looker Studio and Power BI for South African businesses: self-hosted vs per-seat costs, embedding, governance and a plain decision framework.

Read more

More case studies

What Drives the Price of Food? Oil, Fertiliser, El Nino and the 2026/27 Squeeze

An interactive analytical study over decades of public data: how oil, fertiliser and El Nino move South African food prices, what Washington does and does not explain, and a dated forecast for 2026/27.

Read more

From Inverter to Impact: Engineering a Continental Solar-Data Platform

A production data platform that unifies 15 inverter ecosystems, processes global solar irradiance in C++, and surfaces fleet carbon impact and O&M health across 24,350 solar plants.

Read more

Tell me about your business challenge

My offices

  • Johannesburg
    South Africa
    2198