E-Commerce Sales Analytics: Insights & Recommendations

I. Introduction

Background

How can we leverage real-world sales and transaction data to derive insights into consumer trends? Which key insights form the foundational basis for strategic business recommendations? These were the central questions behind my project. In today's digital age, obtaining a multi-dimensional view of sales has become increasingly feasible, thanks to the abundant availability of consumer and seller data. Making minor adjustments to a business model has the potential to cause significant financial implications, sometimes in the millions. A thorough understanding of consumer demands and supplier limitations is necessary for informed decision-making. Consequently, my aim was to uncover the answers to these pivotal questions.

Tools Used

RStudio
Tableau

II. Data

Sources

I have chosen to utilize the dataset provided by Olist, Brazil's largest department store. Olist's recent achievements have been truly exceptional, earning it recognition as one of the fastest-growing startups globally by Growjo in 2022 (Meta). The dataset encompasses information on 100,000 orders placed on their e-commerce platform between 2016 and 2018. This comprehensive dataset comprises nine CSV files, totaling 1,550,922 rows. I performed joins using both Tableau and R-Studio as required to integrate the data effectively. Within this dataset collection, one can find details such as consumer and seller locations, purchase amounts, product categories, product ratings, and additional relevant data.

CUT-DDV

Context: This project is exclusively centered on the sales and performance metrics of Olist, specifically focusing on the Brazilian marketplaces and addresses. The sales data is analyzed from various angles, including product popularity, geospatial distribution, and consumer satisfaction. It is important to note that the visual representations within the project only contain two years of data. Consequently, I searched for potential trends on a monthly or quarterly basis, rather than an annual basis. The project is designed with scalability in mind, meaning it could be easily expanded given a greater quantity of data. The interactive dashboard visualizations are accessible through my website or via Tableau Public, enabling users to dynamically engage with the data and explore insights at their convenience.

User: These visualizations are designed to cater to individuals interested in gaining insights into Olist's sales performance. They also may be particularly valuable to aspiring entrepreneurs looking to start a startup or individuals seeking to gain deeper insights into the Brazilian marketplace.

Task: I aimed to create intuitive and user-friendly dashboards that empower users to formulate and answer their own questions. Because I did not have specific queries from stakeholders, I approached the data analysis from multiple perspectives that could be of interest to potential stakeholders. By offering users a wide range of filtering options, I sought to enhance flexibility in visualization interaction, allowing users to customize their experience and delve deeper into the data based on their unique interests and inquiries.

Data Types: This dataset consists of continuous data types, encompassing variables such as payment value and delivery days. I derived delivery days by calculating the days between an order’s purchase time and delivery date. Ordinal data is represented by the order rankings provided by consumers. Nominal data includes variables such as product category, order ID, consumer ID, and seller ID. Additionally, the dataset contains geographic/spatial data represented by city, state, zip code, latitude, and longitude. Temporal data is captured through order delivery date, order purchase time, etc. To facilitate trend analysis, the order purchase time has been aggregated into monthly and quarterly intervals.

III. Visualization

Demand & Sales Analysis Dashboard

Top 5 Most Purchased Products: Consumer & Seller Info

The main purpose of this dashboard is to understand consumer preferences and analyze purchase volume and revenue trends throughout the course of a year. I determined the top five product categories by totaling the number of orders placed in each category. These categories are Bed, bath, and table, Computers and tech accessories, Furniture and decor, Health and beauty, and Sports and leisure, in order of decreasing popularity. Based on their conventional color associations, each category has been assigned a unique color. The dashboard allows users to filter results based on product category, consumer city, seller city, and quarter of purchase.

Supply Density & Seller Assessment Dashboard

The dashboard aims to analyze supply density and evaluate seller performance. Users can select a specific state to focus on where sellers are located. The primary objective is to assess customer satisfaction and identify any factors that may contribute to variations in satisfaction levels.

IV. Final Thoughts

Ultimately, this project aims to cater to a diverse range of perspectives, maximizing the insights that can be gained from the dashboards. Investigation focus should be customized to match individual interests, whether that involves exploring geographic patterns, analyzing consumer behavior, evaluating seller metrics, considering timing aspects, or examining specialized product categories. It is also worth noting that these dashboards could serve as a tool to help identify variables that are currently absent from the dataset but could contribute to a more in-depth analysis. Given that the dataset spans sales data from 2016 to 2018, regular updates could provide a more comprehensive understanding of ongoing trends and patterns.