Project Introduction and Objective
E-commerce platforms thrive on data. Sellers and marketplaces like Jumia need to understand how pricing, discounts, and customer feedback influence product performance. This project set out to build an interactive Excel dashboard that transforms raw Jumia product data into actionable insights.
The goal was not just to create charts, but to audit, clean, analyze, and visualize the data in a way that supports evidence-based recommendations for sellers.
Dataset and Business Questions
The dataset contained product names, current and old prices, advertised discounts, customer reviews, and ratings. Reviews were used as a proxy for engagement, since sales volumes were not available.
Key business questions:
- Do larger discounts attract more reviews?
- Do highly rated products show stronger engagement?
- Is there a relationship between price and rating?
- Which products perform best, and which need pricing or marketing adjustments?
Initial Data-Quality Audit
Before cleaning, the dataset revealed several issues:
- Misspelled headers (e.g.,
Ratingdinstead ofRating) - Negative review counts
- Blank values in reviews and ratings
- Prices expressed as ranges
- Duplicate product rows
These anomalies were documented in a Data Dictionary worksheet to ensure transparency.
Cleaning and Preparation Decisions
Data cleaning followed clear, traceable rules:
- Prices: stripped of “KSh” and commas, converted to currency (KES).
- Discounts: converted to decimals.
- Reviews: negative signs treated as scraping artifacts, converted to absolute values.
- Ratings: standardized to decimals, blanks left as missing.
- Price ranges: replaced with midpoints, while preserving original text for auditability.
- Duplicates: removed only when all fields matched.
PivotTable and Analysis Workflow
PivotTables powered the analysis:
- Rating Mix → distribution of Poor, Average, Excellent products
- Discount Mix → proportion of Low, Medium, High discounts
- Price vs Rating → average rating by price category
- Engagement by Discount → reviews by discount category
- Top Products → ranked by rating, reviews, and discount
Scatter plots with trendlines and correlation coefficients tested relationships between:
- Discounts vs Reviews
- Ratings vs Reviews
- Price vs Rating
Dashboard Design and Slicer Connections
The dashboard was designed for single-screen readability:
- KPIs row: total products, average price, average discount, average rating, total reviews
- Ranked tables: top products by rating, reviews, and discount
- Scatter plots: discount vs reviews, rating vs reviews, price vs rating
- Mix charts: rating and discount distributions
- Slicers: interactive filters for rating, discount, and price categories
Consistent formatting ensured clarity:
- KES for prices
- Decimals for discounts and ratings
- Thousands separators for reviews
Key Findings
1. Price vs Rating
- Products priced between KSh 1600–1999 show the highest average rating (0.87).
- Lower-priced items (KSh 400–799) also perform well (0.83).
- Mid-range prices dip slightly, showing weaker ratings. ➡️ Both affordable and moderately premium products attract better ratings.
2. Engagement by Discount
- Engagement peaks in the 0.2–0.3 discount band (~18 reviews).
- Very high discounts (>0.5) show lower engagement (~7 reviews). ➡️ Moderate discounts are the sweet spot for driving customer interaction.
3. Top Products by Rating
- Near-perfect ratings (0.96):
- LASA Folding Table Serving Stand
- 40cm Gold DIY Acrylic Wall Sticker Clock
- Portable Home Small Air Humidifier
4. Top Products by Reviews
- 120W Cordless Vacuum Cleaner leads with 69 reviews but has a low rating (0.56).
- Other high-engagement products:
- 137 Pieces Cake Decorating Tool Set (55 reviews)
- Electronic Digital Vernier Caliper (49 reviews)
- 3D Waterproof EVA Shower Curtain (44 reviews) ➡️ High engagement does not always align with high ratings.
5. Top Products by Discount
- 6-in-1 Bottle Can Opener tops with a 64% discount.
- Heavy discounting concentrated in low-cost household goods. ➡️ Sellers use price cuts to stimulate demand in this category.
Correlation Analysis
- Price vs Rating: Weak positive correlation (0.191426). Higher prices show a slight tendency toward better ratings, but the link is weak.
- Ratings vs Reviews: Weak positive correlation (0.14671). More reviews loosely align with higher ratings, but the relationship is minimal.
- Discount vs Reviews: Weak negative correlation (-0.13682). Larger discounts are slightly associated with fewer reviews.
➡️ Overall, none of the variables strongly predict the others.
Business Recommendations
- Test discount bands to optimize engagement without eroding margins.
- Improve listing content for products with visibility but average ratings.
- Investigate quality issues for highly discounted but poorly rated items.
- Promote products that combine excellent ratings with strong engagement.
Limitations and Lessons Learned
- Reviews are a proxy for engagement, not sales.
- Missing values were left blank rather than imputed.
- Price ranges required assumptions (midpoints).
- Correlation does not imply causation — discounts may coincide with other factors like product age or visibility.
Dashboard Screenshot
Jumia Product Dashboard Screenshot
Conclusion
This project demonstrates how Excel can be used not just for reporting, but for data-driven decision-making in e-commerce. By combining careful cleaning, structured analysis, and interactive dashboards, sellers gain practical insights into how pricing and customer feedback shape product performance.
Top comments (0)