ULMS562 Business Analytics for Organisations – Product Insight Report Example

Students enrolled in ULMS562 Business Analytics for Organisations are often required to combine data collection, text analytics, business intelligence, and managerial decision-making into a single project. One assignment that demonstrates these skills is the Product Insight Report, where students collect customer reviews, analyze textual data, and generate evidence-based business recommendations.

This page showcases a sample project involving web scraping, natural language processing (NLP), Python programming, and business analytics techniques.

Assignment Overview

The objective of the project was to analyze customer reviews for a specific consumer product and identify actionable insights for the company. Rather than relying on individual reviews, the analysis examined patterns across a large collection of customer feedback to uncover recurring strengths, weaknesses, and opportunities for improvement.

For this project, 150 customer reviews were collected and analyzed using Python. The goal was to determine what factors drove customer satisfaction and dissatisfaction while developing recommendations that management could realistically implement.

Data Collection and Web Scraping

The first phase involved collecting customer reviews from a publicly available product review platform. Data collection was performed using Firecrawl, a modern web scraping platform capable of extracting information from dynamic websites.

The dataset included:

• Customer review text
• Star ratings
• Review dates

Following data collection, duplicate entries were removed and the dataset was cleaned to prepare it for analysis. The resulting dataset contained 150 unique customer reviews suitable for quantitative and qualitative analysis.

Descriptive Text Analysis

After cleaning the review text, natural language processing techniques were applied to identify the most frequently discussed topics.

A frequency-based document-term matrix was created using Python and Scikit-learn. This allowed the identification of common terms appearing across the review corpus.

The analysis revealed that customers frequently discussed:

• Comfort
• Product sizing
• Overall quality
• Product fit
• Brand-related features

While individual words provided useful information, they did not fully explain customer sentiment. Additional analysis was required to understand the context behind these terms.

Rating-Based Comparisons

To identify the factors driving positive and negative customer experiences, reviews were divided into two groups:

• High-rating reviews (4 and 5 stars)
• Low-rating reviews (1 and 2 stars)

Using proportional text analysis, word usage patterns were compared across both groups.

The findings suggested that satisfied customers frequently emphasized comfort, style, and product satisfaction. In contrast, dissatisfied customers focused more heavily on sizing concerns and product-related issues.

This type of comparison is valuable because it highlights not only what customers discuss, but also which topics influence their overall ratings.

N-Gram Analysis

One limitation of traditional word-frequency analysis is that individual words often lack context.

To overcome this limitation, bigram analysis was performed. Bigrams analyze pairs of words that frequently appear together, allowing more meaningful customer insights to emerge.

Examples of recurring phrases included:

• "True size"
• "Super comfortable"
• "Quality control"

These phrases provided significantly more business value than isolated words because they revealed the specific concerns and experiences customers were describing.

The analysis showed that positive reviews frequently referenced comfort and sizing satisfaction, while negative reviews more often referenced manufacturing and quality concerns.

Custom Dictionary Analysis

A custom business-focused dictionary was developed to measure specific product attributes discussed by customers.

The categories included:

• Comfort
• Sizing and fit
• Product quality
• Style and aesthetics
• Recommendation intent
• Product problems

Each review was scored according to the frequency of terms associated with these categories.

The results revealed that dissatisfied customers discussed quality concerns and product issues substantially more often than satisfied customers. Meanwhile, highly rated reviews contained significantly more references to comfort and aesthetics.

Key Business Insight

The most important finding from the analysis was that customer dissatisfaction appeared to be driven primarily by quality-control concerns rather than product design.

While customers generally appreciated the appearance and comfort of the product, negative reviews frequently mentioned defects, inconsistencies, and manufacturing-related issues.

This insight suggested that operational improvements within the production process could have a meaningful impact on customer satisfaction without requiring changes to the product's design.

Skills Demonstrated

This project demonstrates several skills commonly developed in ULMS562 Business Analytics for Organisations, including:

• ✓ Business Analytics
• ✓ Python Programming
• ✓ Firecrawl Web Scraping
• ✓ Natural Language Processing
• ✓ Text Mining
• ✓ Review Analysis
• ✓ Data Cleaning & Viz
• ✓ Business Reporting

Why This Matters

Modern organizations generate vast amounts of customer feedback every day. Business analysts are increasingly expected to transform this unstructured information into actionable insights that support decision-making.

Projects such as the Product Insight Report provide practical experience in combining data collection, analytics, and business strategy to solve real-world problems.