Sales Performance Analysis Using Business Intelligence Techniques

Abstract

Sales data is only useful when you know how to read it. A business can be growing its revenue while quietly losing margin — and without the right analytical setup, that problem stays invisible until it becomes serious. This project is about building the setup that makes those problems visible before they compound. I am Anandhu A Unnithan, a third-year B.Tech student in Computer Science Engineering (AI & ML) at AISAT, Kochi. My background in data analytics includes a virtual internship with Deloitte through Forage, a data visualisation programme with Tata Group, and hands-on product experience at Deskday Technologies. Through these I have built multiple Power BI dashboards — including a COVID-19 US trends dashboard and a funeral management operations dashboard — that gave me the practical foundation for this project. This report documents a complete Business Intelligence project focused on sales performance analysis. The work covers the full pipeline: ingesting raw sales data, cleaning and transforming it in Power Query, modeling it as a star schema in Power BI Desktop, writing DAX measures for core KPIs, and building three interactive dashboard pages for executive, regional manager, and product analyst users. Key findings include a margin compression problem driven by over-discounting, clear regional profit disparities, and a customer concentration pattern that represents a business risk. The KPI framework and sales performance concepts referenced in this project are drawn from Xoxoday\'s business analytics resources, which provided practical guidance on which metrics carry the most decision-making value in a sales intelligence context.

Introduction

This project explores how sales analytics, developed through a Deloitte Forage internship experience, can transform raw transaction data into actionable business insights using Microsoft Power BI. It highlights the importance of structured analysis over static reporting, emphasizing that organizations often already possess valuable data but lack the analytical framework to interpret it effectively.

Sales performance analysis focuses on understanding business trends across products, regions, customers, and time. Unlike traditional reports, interactive dashboards allow deeper exploration of issues such as profitability, regional performance, customer value, and discount impact. The project is fully built using Power BI, leveraging its ETL capabilities, data modeling, and DAX language to create dynamic and decision-oriented insights.

The dataset used is the Superstore Sales dataset, containing nearly 10,000 transactions from 2020–2023. It includes sales, profit, discounts, customer segments, and geographic information. Significant effort was spent on data cleaning in Power Query, including handling null values, standardizing formats, correcting data types, removing duplicates, and flagging abnormal high-discount transactions.

A star schema data model was designed with a central fact table and supporting dimension tables (date, product, customer, geography, and shipping mode), improving performance and analytical flexibility. Key DAX measures were developed to calculate KPIs such as revenue, profit, margin, YoY growth, average order value, and discount impact.

The dashboard was designed for three user roles:

• Executives: overall business health and trends
• Regional managers: geographic performance and problem areas
• Product analysts: product-level profitability and discount effects

Key findings from the analysis include:

• Revenue is growing, but profit margins are declining due to increased discounting.
• Regional performance varies significantly, with the South underperforming due to specific product losses.
• Technology is the strongest category, while Furniture drags profitability due to low or negative margins.
• A small group of top customers contributes the majority of revenue and profit, indicating high concentration risk.
• Strong seasonality exists, with Q4 (especially November) being the peak sales period.

Conclusion

The most important thing this project reinforced is that business intelligence is primarily about questions, not charts. The charts are the medium. The value is in knowing which questions to ask, building the model correctly so those questions can be answered, and reading the results clearly enough to say something that is actually useful to a business. What the analysis found was not obvious from the headline numbers. Revenue was growing — which looks healthy. But underneath that, margin was eroding because of a discounting pattern propping up volume at the cost of profitability. The South region had a specific product-category problem, not a general sales weakness. The customer base had a concentration risk invisible in aggregate metrics. Those findings are exactly what a properly built dashboard should surface. As a third-year B.Tech student with a focus on AI and ML, I am building this BI foundation deliberately. The most interesting analytical work sits at the intersection of solid data engineering, clear business modeling, and the machine learning I am studying at AISAT. This project is a concrete step in that direction, and the next steps are already defined.

References

[1] Xoxoday Business Analytics Resources (2024). Sales Performance Metrics and KPI Frameworks for Business Intelligence. https://www.xoxoday.com [2] Microsoft Corporation (2024). Power BI Documentation — DAX Reference and Power Query M Language Reference. https://learn.microsoft.com/en-us/power-bi/ [3] Kaggle (2024). Superstore Sales Dataset — Retail Transaction Benchmark. https://www.kaggle.com/datasets/vivek468/superstore-dataset-final [4] Deloitte via Forage (2026). Data Analytics Virtual Internship Programme — Simulation Project Work. [5] Tata Group via Forage (2026). Data Visualisation: Empowering Business with Effective Insights — Certification Programme.

Copyright

Copyright © 2026 Anandhu A Unnithan. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.