Introduction
Data analysis is not just about creating charts; it is about turning raw and imperfect data into useful business insights. This project focuses on building an interactive Excel dashboard to analyze Jumia product data and understand product performance. Because the dataset contained issues such as missing values, inconsistent formatting, and data quality errors, the first step was to clean and organize the information before any analysis could be done. Since sales data was not available, review counts were used as a proxy for customer engagement. This project shows how careful data preparation, categorization, and visualization can help support better decision-making in an e-commerce setting.
Data Quality Audit and Cleaing
Before any useful analysis could begin, the dataset had to be checked carefully for errors and inconsistencies. The raw data contained issues such as missing values, negative review counts, and duplicate entries, all of which could affect the accuracy of the results. To avoid misleading conclusions, the cleaning process focused on preserving data integrity rather than simply making the dataset look tidy.
One important correction was the handling of negative review counts. Since these were not meaningful values in the context of customer engagement, they were converted into positive numbers. Missing values were also treated carefully, because replacing them with zeros could have incorrectly reduced averages and distorted the analysis. In some cases, data such as price ranges was converted into a single usable value by using the midpoint, which allowed the information to remain part of the dataset without losing its meaning.
Duplicate rows were also removed when they represented repeated records, but similar product names with different prices were kept if they appeared to be separate listings. This careful approach ensured that the cleaned dataset was reliable and suitable for further analysis. In short, the data cleaning step was essential because accurate insights depend on accurate input.
Data Enrichment and categorization
After the data was cleaned, the next step was to enrich it by creating new fields that made the dataset more useful for analysis. Raw numbers can be difficult to interpret on their own, so adding categories and calculated values helped turn the data into something more meaningful and easier to compare.
Several useful transformations were made during this stage. Discount values and discount percentages were calculated to give a clearer picture of how much each product was actually reduced. Products were also grouped into categories such as low, medium, and high price levels, as well as poor, average, and excellent rating levels. Since review count was being used as a proxy for customer engagement, products with high review activity were identified separately to show which items appeared to attract more attention. In addition, flags were created for missing ratings or reviews so that incomplete data could be identified easily during analysis.
This enrichment step improved the quality of the analysis because it made the data easier to interpret and compare. Instead of only looking at individual numbers, the project could now reveal broader patterns and trends across product groups.
Analysis and Key Findings
After the data had been cleaned and enriched, the next stage was to analyze it and look for meaningful relationships. The main purpose of this step was to understand whether factors such as discount levels or product prices had any clear effect on customer engagement and ratings. PivotTables and correlation analysis were used to compare the variables and identify possible patterns.
The results showed that some common assumptions were not supported by the data. For example, higher discounts did not strongly lead to more reviews, which means that lowering prices alone does not guarantee more customer attention. There was also no strong relationship between product price and rating, suggesting that more expensive products were not necessarily rated better than cheaper ones.
Another important finding was that some products had high discounts but still performed poorly in terms of engagement or ratings. This shows that discounting alone cannot solve deeper issues such as poor product quality or weak presentation. At the same time, the analysis identified a group of products with both strong engagement and high ratings, which can be seen as examples of successful listings on the platform.
Overall, this step turned the project from simple data preparation into real business insight. It showed that success depends on more than price cuts and that factors like product quality, description, and presentation are likely to have a bigger impact on performance.
Conclusion and Limitation
This project shows that data analysis is most valuable when it is based on careful cleaning, thoughtful assumptions, and clear interpretation. One limitation of the study is that it did not include direct sales, revenue, or product age data. Because of this, review counts had to be used as a proxy for customer engagement rather than a direct measure of sales performance. Some values also required estimation, such as price ranges, and rating groups were created through categorization, which introduced some assumptions into the process.
Another important limitation is that correlation does not mean causation. Even if two variables appear related, it does not prove that one directly causes the other. This means the findings should be interpreted carefully and not treated as absolute proof. Despite these limitations, the analysis still provided useful insights. It suggested that sellers should not depend only on discounts to attract engagement. Instead, they should focus more on product quality, listing presentation, and customer feedback.
In conclusion, the project demonstrates that successful data analysis depends on more than just creating charts. The real value lies in understanding the data, cleaning it properly, and using it to support better business decisions.
Top comments (0)