Data Cleaning & Preparation
Data Cleaning
This is the process of finding and fixing mistakes, duplicates, or missing information in a given data so it is correct and ready to use.I cleaned the dataset by addressing the most glaring inconsistencies.
Check for Missing Values
I inspected all columns for blank or null values.Missing prices were investigated and corrected where possible before converting "Unknown" values in reviews and ratings to blank/null values.
Formulas applied included
COUNTBLANK() for each column.
Apply Filters → Blanks to identify missing records.
Removal of Duplicate Records
To check whether the same product appears more than once;
`- Select entire dataset.
- Go to Data → Remove Duplicates.
- Use Product Name and Price columns as reference fields.`
Duplicate records were removed to ensure each product appears only once.
Standardize Rating Column
I ensured ratings are stored as numeric values between 0 and 5.• Standardized the rating information by extracting the numeric rating from values such as "4.6 out of 5".This was done manually using the Find and Replace functionality.
Convert Reviews to Numeric Format
I converted negative review counts (for example -14.00) to positive values, since review counts should not be negative.
Convert Price Columns to Numeric Format
The assignment requires removing currency symbols and unnecessary characters.Price ranges stored as text instead of numeric values for certain products have been adjusted by determining the averages.
The dataset was cleaned by checking for missing values, duplicate records, incorrect data types, and inconsistencies. Currency symbols, commas, and negative signs were removed from the Current Price and Old Price columns and converted into numeric format. Review counts and ratings were standardised into numerical values to support analysis. Duplicate records were removed and validation checks were performed to ensure ratings ranged from 0 to 5, prices were positive, and discount percentages were reasonable. The resulting dataset was accurate, consistent, and ready for enrichment and analysis.
Data Enrichment.
Is the process of adding extra, helpful details to information you already have to make it more useful.I created the following additional columns to improve my analysis.
1. The Absolute Discount Amount
Formula = [Old Price] - [Current Price]
2. Rating Category
I created a rating category column classified as follows;
| Rating | Category |
| Below 3 | Poor |
| 3 to 4.5 | Average |
| Above 4.5 | Excellent |
Excel Formula
Assuming Standardized Rating is in H2: =IF(J2<3,"Poor",IF(J2<=4.5,"Average","Excellent"))
3. Discount Category
I then created a rating category column classified as follows;
| Discount % | Category |
| Below 20% | Low Discount |
| 20%-40% | Medium Discount |
| Above 40% | High Discount |
Excel Formula
Assuming Discount % is in E2:
=IF(F2<20%,"Low Discount",IF(F2<=40%,"Medium Discount","High Discount"))
4. Price Category
Lastly, I created a price category column classified as follows;
| Current Price (KSh) | Category |
| Less 2,000 | Low Price |
| 2,000-3,000 | Medium Price |
| Above 3,000 | High Price |
Excel Formula
Assuming Current Price is in B2:
=IF(B2>3000,"High Price",IF(B2>=2000,"Medium Price","Low Price"))
To improve analysis, additional calculated fields were added to the dataset. A Discount Amount column was created by subtracting Current Price from Old Price. Products were grouped into Rating Categories (Poor, Average, Excellent) based on customer ratings, Discount Categories (Low, Medium, High) based on discount percentages, and Price Categories (Low, Medium, High) based on current selling prices. These derived fields simplified segmentation, enabled trend analysis, and supported dashboard visualisations.
Data Analysis
1. Descriptive Statistics
Current vs Old Price: The average current price (KSh 1,187) is substantially lower than the average old price (KSh 1,811), indicating that products are generally offered at discounted prices.
Discount Level: The average discount of 37% suggests that the marketplace applies fairly aggressive promotional pricing, making products attractive to price-sensitive customers.
Customer Satisfaction: The average rating of 3.9 out of 5 indicates overall positive customer feedback, though there is room for improvement in product quality or customer experience.
Product Range: With 112 products, the dataset provides a reasonably diverse sample of items across different categories.
Review Activity: A total of 723 reviews shows moderate customer engagement, providing sufficient feedback for assessing product performance.
Price Spread: The large gap between the least expensive product (KSh 38) and the most expensive product (KSh 3,750) indicates a wide range of product affordability, catering to different customer segments.
2. Trend Analysis
A. Does a Higher Discount Percentage Result in More Customer Reviews?
Based on the products with available review data, there is no strong positive relationship between discount percentage and average number of reviews.
While some highly discounted products attract many reviews, many others do not. Discount percentage alone does not appear to drive customer review volume.
B. Do Highly Rated Products Receive More Customer Reviews?
The evidence suggests a weak relationship between ratings and review counts.
Highly rated products do not necessarily receive more reviews. Review count appears to depend more on popularity and sales volume than on ratings.
C. Are Expensive Products Rated Higher Than Cheaper Products?
The dataset shows no consistent relationship between price and rating.
Expensive products are reasonably rated higher, while cheaper products edge out medium priced products. Product quality and customer satisfaction appear to have a stronger impact on ratings than price.
D. Top 5 Highest-Rated Products
Several products achieved the maximum rating of 5.0/5. The first five are:
E. Top 5 Lowest-Rated Products
These products recorded the lowest customer satisfaction scores in the dataset.
Product Performance Analysis
A. Top 10 Products with the Highest Discounts
Most of the highest-discounted products are low-priced household and decorative items. The highest observed discount is 64%, indicating aggressive promotional pricing.
B. Top 10 Products with the Highest Number of Reviews
Cleaning tools, home improvement products, and crafting kits generate the highest customer engagement, attracting significantly more reviews than most other products.
C. Top 10 Highest-Rated Products
Most of the top-rated products are practical household items that combine affordability with functionality.
D. Products with High Discounts but Low Ratings
These products present a potential quality or customer satisfaction concern.
Large discounts do not necessarily indicate high product value. Several heavily discounted products received poor customer ratings, suggesting customers may have been dissatisfied with quality or performance.
E. Products with Strong Customer Engagement
Strong engagement can be identified by combining high review volume and good ratings.
Seller Performance Analysis
A. Which Products Appear to Have Strong Customer Demand?
Products with both high review counts and good ratings are likely experiencing strong customer demand.
These products demonstrate a combination of:
- Strong customer interest (high review volume).
- High satisfaction levels (ratings above 4.0).
- Good market acceptance.
They should remain highly visible in promotions because they already perform well.
B. Which Products May Require Better Pricing or Marketing Strategies?
Products with few reviews despite receiving substantial discounts may not be attracting sufficient customer attention.
These products are well-rated and heavily discounted but still attract very few reviews. This may indicate:
- Low product visibility.
- Ineffective marketing.
- Limited customer awareness.
- Niche market demand.
Increasing advertising exposure could improve sales performance more effectively than offering additional discounts.
C. Are There Products Receiving Many Reviews but Having Average Ratings?
Yes. Several products receive substantial customer attention but achieve only average satisfaction scores.
The most notable example is the 120W Cordless Vacuum Cleaner, which generated the highest engagement in the dataset (69 reviews) but received a relatively poor rating (2.8/5).
This suggests:
- Strong demand and visibility.
- Product quality or performance issues.
- Potential risk of negative customer perception if improvements are not made.
D. Are There Products with High Discounts but Low Customer Engagement?
Yes. Some products have very attractive discounts but attract little customer interaction.
Business Insights
- Customer engagement is influenced more by perceived value and product quality than by discount size.
- Highly rated products should receive greater promotional support to increase visibility and sales.
- Products with high engagement but poor ratings require quality improvements to protect customer satisfaction and brand reputation.
- Data-driven pricing strategies are likely to be more effective than broad discounting campaigns.
Recommendations
- Focus on product quality improvements for products receiving low ratings despite high sales activity.
- Increase visibility of highly rated products through sponsored listings, featured placements, and targeted promotions.
- Maintain adequate stock levels for products demonstrating strong customer demand.
- Adopt strategic discounting rather than relying on large discounts to drive engagement.
- Monitor customer reviews continuously to identify product weaknesses and opportunities for improvement.
Conclusion
The analysis shows that customer satisfaction, product quality, and visibility are stronger drivers of success than large discounts. Sellers who focus on delivering quality products, maintaining positive customer experiences, and using targeted marketing strategies are likely to achieve better long-term performance on Jumia.
Top comments (0)