Introduction
When I first looked at the Jumia product dataset for this project, it seemed fairly simple.There were product names,prices,discounts,reviews and ratings. At first glance, it looked like a dataset that could easily be summarized with a few Excel formulas and charts.
But after opening the data and actually trying to analyze it, I quickly realized that the difficult part was not creating the charts. The real challenge was getting the data into a form that Excel could actually analyze correctly.
Some prices contained currency symbols,ratings were written as text such as "4.5 out of 5" and the review data needed cleaning before it could be used confidently. This made the project a good practical example of what happens in a real data analysis process.
For this project, I used Microsoft Excel to clean and analyze Jumia product data and then created an interactive dashboard to explore product pricing,discounts,customer reviews and ratings.
Understanding the Dataset
The dataset contained information about 115 Jumia products.
The original columns were;
- Product
- Current Price
- Old Price
- Discount
- Review
- Rating
These variables provided enough information to explore product pricing and customer engagement.
The Current Price represented the selling price of the product while Old Price represented the price before the discount. The Discount column showed the percentage reduction, while Review and Rating provided information about customer engagement and satisfaction.
Raw data before cleaning
Data Cleaning and Preparation
Before starting the analysis, I first cleaned the dataset to make sure the information was consistent and suitable for analysis. I focused on several issues I identified in the original Jumia data.
Removing Duplicates
I first checked for duplicate products using Conditional Formatting in Excel. This made it easier to visually identify repeated records before removing them.
Standardizing Text
I standardized the text in the product column to make the product names more consistent.I did this by highlighting the columns then right clicking and then formating them to numbers. This is important because small differences in text formatting can make the same product appear as different values during analysis.
Removing Negative Values
I also checked the review column for negative values and corrected the invalid entries byFind and replace then standardised the numericals.
Standardizing the Rating Column
The rating column was originally stored as text, with values such as_ 4.5 out of 5_. I converted these values into numerical ratings.
I extracted the numerical part using:
Ctrl+H find out of 5 and replace with blank
This converted a value such as 4.5 out of 5 into the numerical value 4.5.
Correcting Rating Spelling
I also corrected inconsistencies and spelling issues in the rating related data so that the ratings followed one standard format.
Standardizing Prices
The price columns also needed cleaning because some prices contained currency symbols, commas, or inconsistent formatting.
For example:
KSh 1,525
was converted into a numerical value that Excel could use for calculations.Then converted the columns to currency and used Kes
This allowed me to calculate statistics such as average price, minimum price, maximum price and discount amounts.
Cleaned price columns
Creating New Fields
After cleaning the original columns, I created additional fields to make the analysis more useful.
1.Discount Amount
I created a Discount Amount column by subtracting the current price from the old price:
Discount Amount = Old Price − Current Price
For example:
KSh 1,525 − KSh 950 = KSh 575
2.Rating Categories
I grouped products into three rating categories,
Poor: Below 3
Average: 3–4
Excellent: Above 4.5
=IF(I3<3,"Poor",IF(AND(I3>=3,I3<=4),"Average",IF(I3>4,"Excellent")))
This made it easier to compare groups of products and later use them in PivotTables and dashboard slicers.
3.Discount Categories
I also categorized discounts as,
Low Discount: Below 20%
Medium Discount: 20%–40%
High Discount: Above 40%
=IF(G3<20%,"Low discount",IF(AND(G3>=20%,G3<=40%),"Medium discount",IF(G3>40%,"High discount")))
4.Price Category
I created this field containing Low,Medium and High Price categories. This was particularly useful when creating the dashboard filters.
=IF(B2<1000,"Low price",IF(AND(B2>=1000,B2<=3000),"Medium price",IF(B2>3000,"High price")))
Descriptive Analysis
Once the data was cleaned, I calculated some basic statistics to understand the dataset.
The average current price was approximately KSh 2,355, while the average old price was approximately KSh 3,575.
The average discount was approximately 36.8%.
The minimum current price was KSh 38, while the maximum was approximately KSh 132,932.
The large difference between the minimum and maximum prices showed that products on the platform covered a very wide price range.
These statistics became the basis for the KPI section of my dashboard.
Exploring Relationships in the Data
After looking at the overall statistics, I wanted to investigate relationships between different variables.
1.Discount vs Reviews
The first question I explored was,
Do higher discounts lead to more customer engagement?
I compared discount percentages with the number of reviews.
The results showed that having a large discount did not automatically mean that a product had many reviews. Some highly discounted products had relatively low review counts, while some products with smaller discounts still received significant customer engagement.
This suggests that discounts can attract attention but they are not the only factor influencing customer engagement.
Product quality, usefulness, price, visibility and customer experience can also play an important role.
2.Rating vs Reviews
This comparison was particularly interesting because the two variables measure different things.
A review count can indicate customer engagement, while a rating can give an indication of customer satisfaction.
For example, the 120W Cordless Vacuum Cleaners Handheld Electric Vacuum Cleaner had the highest review count at 69 reviews, but its rating was only 2.8.
In contrast, the 137 Pieces Cake Decorating Tool Set Baking Supplies had 55 reviews and a rating of 4.6.
A product with many reviews is not necessarily highly rated and a product with a perfect rating may have very few reviews.
3.Price vs Rating
I also compared current prices with product ratings.
The purpose was to find out whether more expensive products necessarily received better ratings.
The analysis did not indicate that price alone determines customer satisfaction.
Highly rated products appeared at different price levels, suggesting that factors such as product quality and perceived value may be more important than price alone.
Product Performance Analysis
The next step was to identify products that stood out based on discounts,reviews and ratings.
Top Discounts
The highest discount recorded was 64%.
Some of the products with the highest discounts included:
6 In 1 Bottle Can Opener Multifunctional Easy Opener- 64%
Creative Owl Shape Keychain Black- 61%
5-PCS Stainless Steel Cooking Pot Set With Steamed Slices- 55%
LASA 3 Tier Bamboo Shoe Bench Storage Shelf- 54%
However, having the highest discount did not automatically mean that these were the best-performing products.
This was one of the reasons I decided to compare discounts with ratings and reviews rather than looking at discounts alone.
Top Reviews and Ratings
The product with the highest recorded number of reviews was the 120W Cordless Vacuum Cleaners Handheld Electric Vacuum Cleaner, with 69 reviews.
The dataset also contained several products with ratings of 5.0.
However, I did not consider a 5.0 rating alone enough to determine overall product performance. A product with a perfect rating but very few reviews may not provide as much evidence as a product with a slightly lower rating and dozens of reviews.
High Discounts and Low Ratings
Another analysis I found useful was identifying products that combined high discounts with low ratings.
This is important because a seller might assume that reducing a product's price will solve poor performance.
However, if customers are giving the product low ratings, the underlying problem may have more to do with product quality,expectations,packaging or the accuracy of the product description.
For example, the 5-PCS Stainless Steel Cooking Pot Set With Steamed Slices appeared among highly discounted products while also having a relatively low rating.
Instead of simply increasing the discount, a seller could investigate customer feedback and determine what is causing dissatisfaction.
Building the Interactive Dashboard
After completing the analysis, I brought the most important findings together into an interactive Excel dashboard.
The dashboard included KPI cards for the main metrics, charts for product performance,category breakdowns and slicers.
The slicers allowed the dashboard to be filtered according to:
_- Rating Category
- Discount Category
- Price Category_
This made the dashboard more interactive because users could focus on particular groups of products instead of viewing the entire dataset at once.
Key Insights
After completing the analysis, several key insights stood out.
First, higher discounts did not automatically result in higher customer engagement. Discounting can help attract customers, but it is not enough by itself.
Second, ratings and reviews should be considered together. A product with many reviews and a strong rating gives a better indication of both engagement and satisfaction.
Third, price does not necessarily determine customer satisfaction. Highly rated products existed at different price levels.
Finally, products with high discounts and low ratings deserve further investigation rather than simply receiving bigger discounts.
Recommendations for Jumia Sellers
Based on the analysis, I would recommend that sellers,
1.Use discounts strategically
Discounts should be combined with good product quality,accurate descriptions,attractive images and good customer service.
2.Monitor reviews and ratings
Sellers should regularly monitor both review volume and ratings instead of relying on one metric.
3.Investigate low-rated products
Products receiving poor ratings should be investigated to understand the cause of customer dissatisfaction before additional promotions are introduced.
4.Review pricing based on performance
Not every product needs the same discount. Pricing strategies should take into account customer engagement,ratings,competition and product value.
5.Use customer feedback
Reviews provide useful information about product weaknesses and opportunities for improvement.
Conclusion
This project helped me understand that Excel can be much more than a tool for entering data and performing simple calculations.
I started with a raw Jumia dataset and used Excel to clean the data, convert text into usable numerical values,create calculated fields, categorize products,analyze relationships and identify product performance patterns.
I learned the importance of questioning the data. Finding unusual values such as an invalid rating reminded me that even a well designed dashboard can produce misleading results if the underlying data is not properly validated.
Overall, the project gave me practical experience with data cleaning, Excel formulas,Pivot tables,Pivot charts,slicers and dashboard design.
More importantly, it showed me how to move from raw data to data that can actually be understood and used.
Top comments (0)