jumia

PHASE 1: Data Cleaning

​We had to scrub the raw data before it could be visualized, and the reason was that we needed to transform raw text into mathematical numbers. Here is my technical process:
​De-Noise: Find and Replace (Ctrl + H) was used to remove Noise: ‘KSh’ and commas in the columns containing prices.
​Normalizing Ranges: There were products with price ranges (e.g., 1,620 - 1,800). I mean, average these ranges to make us have only one numerical point in our charts.
​Standardizing Ratings: Ratings were indicated to be 4.5/5. Text-to-Columns (use space as delimiting) was used to avoid all but the numeric 4.5.
​Absolute Values: The negative numbers appeared in the column of raw reviews. The =ABS() function was used to make sure that all the reviews were positive numbers.
​Deduplication: I have found and deleted 3 duplicates to make sure that our averages are not distorted by the presence of duplicates.

PHASE 2: Data Enrichment

​Raw data finds out what has happened; enriched data is what found out where to look. To fuel our dashboard filters, I have added new logic to the dataset:
​Categories of Ratings: I have used a new column based on an IFS formula to categorize products in:
​Excellent: 4.0 – 5.0
​Good: 3.0 – 3.9
​Needs Attention: Below 3.0
​Discount Tiers: I grouped the discounts into High (> 40), Medium (20-40), and Low (< 20).
​Pricing Segments: I included a column of price ranges (i.e., under 1k, 1k-2.5k, above 2.5k) to enable a manager to restrict by budget.

PHASE 3: Analysis of Data

​I conducted a statistical deep dive after cleaning and enriching. The following are the three Golden Insights that I discovered:
​The Quality Driver: Ratings and Reviews 0.52.
​Meaning: When the products are of high quality, they tend to have more interaction. The most successful thing about Jumia is quality.
​The Discount Myth: I discovered that discounts and reviews had a correlation of 0.01.
​Theory: Customer engagement is not necessarily fuelled by deep discounts. Customers do not trust low-price products that do not have reviews.
​The Value Trap: I have found such products as the 5-piece cooking pot set that has a huge discount of 55 percent and a terrible rating of 2.1.
​Meaning: Big discounts on these products are used to conceal low quality, thus resulting in large returns and dissatisfaction by the customers.

PHASE 4: Dashboard Design

​I designed the dashboard as a one-stop shop of the marketing team of Jumia. It is constructed in three separate districts:
​The KPI Tile Header: Four large numbers dynamically changing: The number of products sold (112), the average rating (3.89), the average discount (37%), and the quantity of reviews (723).
​The Trend Zone: I have added two Scatter Plots. One illustrates Discount vs. Reviews (to establish the myth of the Discount) and the other Rating vs. Reviews (to establish the myth of Quality is King).
​Interactive Slicers: I have included three slicers on the left side (Ratingg Category, Discount Category, and Price Range).
​Functionality: When the Rating Slicer is clicked on the item 'Needs Attention', Jumia immediately displays the products with high discounts, but of low quality.
​Final Recommendations of the Lead Analyst.
​My analysis shows that I can recommend Jumia to:
​Judge Rating more than Price: Within the search results, you should place more emphasis on items with a 4.5 or higher rating than on items with high discounts.
​Fix the Ghost Products: In the 58 products that people have 0 reviews, you should provide the incentive of loyalty to the first few people who review the product to build the early trust.
​Check the 'Value Traps': an item that has a higher discount of over 40 percent and rating of less than 3.0 must be put under the red flag of being checked up with the vendor.