Building an Interactive Excel Dashboard for E-commerce Product Analysis: A Case Study of Jumia Products

Introduction

E-commerce platforms like Jumia generate a constant stream of data on pricing, discounts, and customer feedback, but that data is only useful once someone actually digs into it. For this project, I worked with a dataset of 111 products listed on Jumia to understand how price, discounting, and customer reviews relate to one another, and to turn that understanding into something Jumia's sellers could actually act on.

This article walks through the full process of cleaning a messy real-world dataset, building out the analysis with Excel formulas and Pivot Tables, constructing an interactive dashboard, and the business insights that came out the other end including a couple of assumptions that didn't survive contact with the actual numbers.

Dataset Description

The dataset included six core fields per product:

Product - product name
Current Price - selling price in KES
Old Price - price before discount
Discount - percentage discount applied
Review - number of customer reviews
Ratings - average customer rating out of 5

Before any of this was usable, it needed cleaning.

Data Cleaning and Preparation

The original dataset

Two issues stood out immediately:

The Review column was entirely negative, values like -2, -14, and -69, which makes no sense for a count of anything. This pointed to a sign error somewhere upstream in how the data was collected or exported, and was corrected to positive values.
One row's Current Price was stored as text and range "1,620 - 1,980", a price range typed into a single cell instead of a number. This one bad cell was enough to silently break SUM, AVERAGE, and correlation calculations across the whole column unless it was fixed.

I also had to think carefully about a more subtle issue: roughly half the dataset that is 57 of 111 products, had no Review or Rating data at all. Rather than ignore this gap, I treated it as a finding in its own set. Any conclusion about "what makes a product well reviewed" in this analysis only applies to the half of the catalog that actually has review data.
The empty fields

Enrichment: Calculated Columns

With the base data clean, I added:

Discount Amount = Old Price − Current Price
Rating Category: Poor (below 3), Average (3–4), Excellent (4.5 and above)
Discount Category: Low (below 20%), Medium (20–40%), Above/High (over 40%)
Price Category: Low, Medium, High, based on where a product's price falls within the overall range

The Cleaned and Enriched Dataset

Excel Techniques and Analysis
Descriptive statistics across the 111 products:

Total products 111
Average current price KES 1,181
Average discount 37%
Average rating 3.88
Total reviews 721

I did Correlation analysis, using =CORREL() to test three commonly assumed relationships:

1. Discount % vs. Review -0.14 No relationship
2. Rating vs. Reviews +0.07 No relationship
3. Price vs. Rating +0.01 No relationship

One thing I want to highlight about the Price vs. Rating result specifically, because it changed how I read the rest of the analysis. When I first built a Pivot Table comparing average price across Poor/Average/Excellent rating groups, the averages climbed 998, 1,369, 1,381 which looked like evidence that pricier products get rated higher. But the correlation was 0.01, essentially zero. Building a scatter plot settled it, ratings of 2.0 through 5.0 were scattered across the entire price range, with no visible upward drift. The Pivot averages were misleading because the Poor and Average rating groups only had 12 and 22 products respectively, which are small enough that a couple of outlier prices could swing the average without reflecting any real underlying pattern. It was a good reminder that a summary average can tell a different story than the full spread of the data, and it's worth checking both before drawing a conclusion.

Ranking and segmentation, I used SORT, INDEX+SEQUENCE, and XLOOKUPto build Top-10 lists (by Rating, Reviews, and Discount), and FILTER combining AND (*) and OR (+) logic, to isolate specific product segments.

1. Strong customer demand (Reviews ≥ 20 AND Rating ≥ 4): 9 products, led by a 137-Piece Cake Decorating Tool Set (55 reviews, 4.6 rating).
2. High discount but low rating (Discount ≥ 40% AND Rating < 3): 10 products, including a 5-PCS Stainless Steel Cooking Pot Set discounted 55% but rated only 2.1.
3. High discount but low engagement (Discount ≥ 40% AND Reviews ≤ 3): 9 products — heavily discounted, but barely bought or reviewed.
4. Products needing better pricing or marketing strategies (Reviews ≤ 3 OR Rating < 3): 23 products in total.

Category breakdowns via Pivot Tables and PivotCharts

Rating Category: 22 Average, 22 Excellent, 12 Poor, 55 Not Provided
Discount Category: 57% Above, 21% Medium, 19% Low Discount (approximate)
Price Category: 50% Low, 43% High, 7% Medium

Dashboard Creation Process

The final dashboard pulls all of this into one sheet, organized into four sections:

1. Overview/KPI cards - Total Products, Average Price, Average Discount %, Average Rating, Total Reviews, styled as five distinct summary cards.
2. Product Performance - three bar charts for Top 10 by Rating, by Reviews, and by Discount.
3. Trend Analysis - three scatter charts (no connecting lines, so the actual spread of the data is visible rather than hidden inside a single line or average), each labeled with its correlation value directly in the title.
4. Product Categories - three pie charts (Rating, Discount, and Price Category), connected via slicers so that clicking a single category, instantly re-filters all three pie charts.

Key Insights and Business Recommendations

Are higher discounts leading to higher customer engagement? No. The correlation is -0.14, and the scatter plot backs this up with no visible trend across the full 0–70% discount range.

Do highly rated products have higher or lower prices? Essentially neither. The correlation is 0.01. This was the most counter-intuitive finding of the project, especially since an initial Pivot Table look made it seem otherwise until I checked the raw scatter of the data.

Which products are performing best based on reviews and ratings? The 9 "strong demand" products sheet, consistently high reviews paired with strong ratings, are the most validated products in the catalog and worth featuring or promoting.

**Which products may need improved pricing strategies? **The 23 flagged products split into two genuinely different problems; some like a DIY File Folder rated 5 with only 1 review, simply aren't being seen, which is a marketing or visibility issue. Others like the discounted Cooking Pot Set above, have a real satisfaction problem that a bigger discount won't fix.

Recommendations for Jumia sellers:

1. Don't treat discount depth as an engagement lever - the data shows no payoff from deeper discounts on their own.
2. Close the data gap - nearly half the catalog has no review or rating at all, which limits how much sellers and buyers alike can trust any single product's track record.
3. Audit the 9 "high discount, low engagement" products specifically - Since the discount clearly isn't the barrier, so something else likely is.
4. Don't lump visibility problems in with quality problems - a 5-star product with one review needs marketing. A heavily discounted product with a 2-star rating needs a quality fix, not more discounting.

Closing Thoughts

The most useful part of this project wasn't the dashboard itself, it was the number of instinctive assumptions that "bigger discount = more reviews," and "pricier = better rated" that turned out not to hold up once actually tested. Building the habit of double-checking a Pivot Table average against a correlation and a scatter plot, rather than trusting the first number that looks convincing, was my biggest practical lesson from this project.

The full workbook which has raw data, cleaned data, all analysis sheets, Pivot Tables, charts, and the final interactive dashboard, is available in my GitHub repository linked below.

GitHub repo: [https://github.com/mimi580/Jumia-Product-Performance-Dashboard-An-E-commerce-Data-Analysis-Project/blob/main/Excel%20Jumia%20Dataset.xlsx]