E-Commerce Business Transaction Dashboard

Overview

This dashboard is built using a year's worth of transaction records from an online retailer, it takes a raw spreadsheet of over half a million rows and turns it into a clear, interactive report that anyone on a business team can actually use, not just people comfortable digging through spreadsheets.

Looking at the finished dashboard, a few patterns stood out right away. Sales peaked sharply in November, tied to seasonal demand, while February marked the slowest month of the year, a pattern any retailer in this space would want to plan around rather than be caught off guard by. One product in particular, a popcorn holder, came out on top both in total revenue and in the number of units sold, making it the clearest standout performer in the entire catalog. A more unusual finding involved two specific products that kept appearing at the top of two very different lists at once, both as strong sellers and as the most frequently cancelled items. That kind of overlap doesn't happen by chance, and it usually points to a specific, fixable issue such as inconsistent stock levels or a quality concern with those particular items, rather than a general problem across the business. On the customer side, a small group of buyers accounted for a large share of total revenue, and the vast majority of sales came from a single country, leaving other markets with clear, largely untapped potential.

Based on these findings, a few practical steps stood out. The two products driving repeated cancellations are worth a closer look at their supply chain and quality control, since fixing even one of them could meaningfully reduce lost sales. Inventory planning ahead of the September to November period would also help the business avoid running short during its busiest and most profitable stretch of the year. On the customer side, the small group of high-value buyers responsible for most of the revenue deserves a more deliberate retention approach, whether that's personalized outreach, loyalty incentives, or simply closer account management. At the same time, the fact that international markets already show real demand, despite receiving little attention so far, makes a strong case for testing targeted marketing or better delivery options in a few of those regions. None of these recommendations call for a major overhaul, just a more focused use of information the business already had sitting in its data all along.

Most businesses already have the data they need sitting somewhere in their systems, but rarely in a form that's easy to act on. Numbers get buried across different tabs, formats are inconsistent, and important details like cancelled orders or missing customer records often go unnoticed until they start causing real confusion down the line. Before any dashboard gets built, that raw data gets properly cleaned and structured first, since a polished-looking report built on messy numbers is often worse than no report at all.

In this particular case, the finished dashboard made it possible to see exactly which months brought in the most sales, which products were performing well versus which ones were quietly causing repeat cancellations, and where the bulk of revenue was actually coming from in terms of both customers and countries. These are the kinds of questions that are difficult to answer by scrolling through raw spreadsheets, but become immediately obvious once the data is visualized properly.

This is the value a good dashboard brings to a business: it shortens the distance between having data and making a decision. Instead of waiting on someone to manually pull numbers together every time a question comes up, teams can look at a live, filterable report and get answers in seconds. For companies sitting on years of unused sales, customer, or operations data, this kind of transformation often uncovers issues and opportunities that were there all along, simply waiting to be seen clearly.

Note: For demonstration purposes, this dashboard is built on a synthetic dataset and does not represent real data.

Raw Data

The dataset used for this project contained transaction-level sales records spanning roughly one year, from December 2018 to December 2019, totaling 536,350 rows. Each row represented a single item within a customer's order, captured across eight columns: Transaction Number, Date, Product Number, Product Name, Price, Quantity, Customer Number, and Country. Together, these fields covered 3,768 unique products, 4,738 unique customers, and orders coming in from 38 different countries, giving a broad view of both product performance and customer behavior across markets.

Dashboard