What Does Your Customer Data Really Tell You? | Insights from Online Food Delivery Data

By abhishek.verma75000 · September 14, 2026

Customer behavior is not just about how many orders are placed. The real story is hidden in the data who the customers are, what patterns they show, and…

The analysis identified the pin code areas with the highest customer concentrations along with their positive Output counts. A bar chart was generated comparing the top 10 pin codes side-by-side, showing both total customer volume and positive Output for each area, making it easy to spot which locations drive the most business and successful outcomes.

The 18-20 Female segment shows the highest positive Output rate at 100% (though based on a small sample of 4). Among larger groups, 21-23 Females (90.4%, n=73) and 21-23 Males (90.0%, n=80) lead with strong, well-supported positive rates.

Feedback varies noticeably across Monthly Income levels. Customers with 'No Income' give the most positive feedback (90.9%), followed by 'More than 50000' (80.6%) and '10001 to 25000' (77.8%). In contrast, those earning 'Below Rs.10000' have the lowest positive feedback rate at 56.0%, meaning nearly 44% of this group reported negative feedback - the highest negative rate among all income groups. Two charts were created: a grouped bar chart showing actual customer counts by income and feedback type, and a stacked percentage chart showing the feedback split within each income level.

Regular customers deliver the best outcomes overall, with a 78.9% positive Output rate and an 82.1% positive Feedback rate, resulting in a combined score of 80.5%.

The analysis explored how Educational Qualifications and Marital Status relate to customer Feedback ratings. Three visualizations were generated: a bar chart comparing feedback distribution across education levels, a bar chart comparing feedback distribution across marital status groups, and a heatmap showing the combined positive feedback percentage for each education and marital status pairing. These charts let you see at a glance which groups tend to give more positive or negative feedback, and how the two factors interact together.

Occupation and family size both show clear links to positive outcomes. Students have the highest positive output rate at 88.9% (out of 207 cases), while Self-Employed individuals have the lowest at 63.0%. For family size, households of 4 have the highest positive rate (85.7%), while households of 6 have the lowest (62.1%). Overall, 77.6% of all cases resulted in a positive outcome.

Your dataset was broken down by Gender and Marital Status combined, and a grouped bar chart along with a data table have been generated to show the count of customers in each segment (like Single Male, Married Female, etc.).

There's a weak positive correlation (0.170) between family size and age, meaning family size only slightly tends to increase with age. A scatter plot with a trendline visualizes this relationship, and a bar chart shows the average age and number of records for each family size group, helping to reveal patterns in household composition.

The dataset shows customers have a mean age of 24.63 years and a median of 24.00 years, with a standard deviation of 2.98 years. The distribution is right-skewed (skewness of 0.810), meaning there are more younger customers with a tail extending toward older ages. Two histograms were generated: one showing the overall age distribution in 5-year bins with mean and median markers, and another comparing age distributions by gender to visualize any differences in age profiles between male and female customers.

Using distance from the centroid of all delivery coordinates, 12 out of 388 customer locations were flagged as geographic outliers. The centroid was calculated at latitude 12.9721, longitude 77.6002, with outliers identified using both a z-score method ( z 3) and an IQR method (distance beyond 0.1356 degrees). A map visualization highlights these anomalous locations in red against the main cluster of normal locations in blue, with the centroid marked as a green star. A histogram also shows the distribution of distances from the centroid with the IQR threshold marked, making it easy to see how far the outliers deviate from the typical delivery area. Detailed tables list the specific outlier locations and their distances for further review.