Instacart Customer Segmentation
Instacart is an online grocery store that operates through a mobile app, and they want to learn more about their sales patterns. Stakeholders are interested in the variety of customers and their purchasing behaviors.
-
Description text goes here•Identify peak order hours and days
Analyze customer purchasing behavior, customer loyalty, regional differences, age, income, other key factors.
Determine the most ordered products in various demographic groups to generate targeted marketing campaigns
-
Python
Data wrangling
Deriving new variables
Grouping data
Aggregating data
Reporting in Excel
Population flows
-
Open-source data sets from Instacart (via Kaggle)
Fictional customer dataset from CareerFoundy
-
Python
Excel
Peak Order Times
The first request was to determine when customers were placing the most orders. On the right are two graphs that show orders broken down by hour and by day of week.
The graph on the left shows the busiest days of the week to be Saturday and Sunday (0 represents Saturday, 1 represents Sunday, etc.). The graph on the right shows the busiest hours of the day to be between 9 AM and 4 PM.
This means that people are placing orders mostly on the weekends, and during the hours of a typical work day.
Most Ordered Departments
Next, stakeholders wanted a breakdown of which products are being ordered, and the prices of these most ordered products.
The chart on the right shows that people use Instacart to order produce (as indicated by the legend table to the right of the chart) at almost double the rate of the next highest department, dairy and eggs. The second tier of departments consists of snacks, beverages, frozen items, and pantry staples, but those are a distant second to produce, dairy, and eggs.
In one of the biggest surprises in this analysis, customers are more likely to order perishable items from Instacart by a fairly significant margin.
Customer Loyalty
Next, I wanted to explore the spending and order patterns of customers based on how often they used Instacart to order groceries. The chart below shows the breakdown of customers based on their loyalty status, defined here by the following criteria:
Loyal Customer = 40 orders or more
Regular Customer = 10-40 orders
New Customer = 10 orders or less
Once loyalty status has been established, the next step is to analyze the spending patterns within each group. The charts below illustrate total spending by loyalty status, as well as average spending per loyalty category.
Total spending aligns with expectations; the “regular customer” segment consists of a larger number of individuals, leading to higher overall spending. However, the real insight emerges when examining average spending by loyalty status. New customers spend, on average, over two dollars more per order compared to loyal customers.
Further analysis is required to assess the frequency of orders placed by these customers. This finding lays the groundwork for deeper investigation into the reasons behind the higher per-order spending of new customers, contrasted with the frequent but lower spending habits of loyal customers.
Regional Order Analysis
Analyzing the order data by region reveals that the Southern region of Instacart has the highest total revenue, comprising 38% of the customer base. This region demonstrates a marked preference for grocery expenditures, leading to a collective total spending of approximately 130 million dollars.
The Midwest follows as the second highest region, with total expenditures just below 100 million dollars. In terms of individual spending patterns, Midwest customers stand out by spending an average of $12.20 per order, surpassing the overall average of $11.67.
Although the average spending per order across regions remains relatively consistent, the total spending visualization indicates a significant disparity. This results in a notable gap of 30 million dollars in overall spending between the Southern and Midwest regions.
The amounts on the y-axis of the visualization are represented in increments of one hundred million dollars, further emphasizing the substantial financial engagement of the Southern customer base compared to their Midwestern counterparts.
Customer Segmentation
For this particular analysis, I chose to categorize the customer base based on dependency status—specifically, differentiating between parents and non-parents—as well as income levels. Additionally, I further segmented each group by age, resulting in a total of fifteen distinct clusters for the analysis.
Dependent-Based Clusters
Parents are power users: Parents spend significantly more than customers without dependents, with middle-aged parents driving $136.1M in total spending, over 6x more than their non-parent counterparts.
Consistent order value: While parents spend more in total, the average order value remains relatively consistent across segments (~$11.90-$12.40).
Strong parent loyalty: Parents order slightly more frequently (every 11.1 days) compared to those without dependents (11.2 days).
Income-Based Clusters
Middle-income dominance: Middle-income customers account for the largest portion of total spending ($244.3M), particularly Middle-Income Middle-Age Adults ($112.4M).
High-income ordering frequency: High-income customers across all age groups place orders more frequently (every 10.3 days on average) compared to middle and low-income segments.
Premium product opportunity: High-income segments show dramatically higher average order values ($250-$270 vs $7-$13), highlighting an opportunity for premium product targeting.
Marketing Implications
The analysis identified three high-value customer segments that represent the greatest ROI potential for targeted marketing efforts:
Middle-Aged Parents: Focus on family-oriented promotions and bulk purchasing options to increase basket size.
High-Income Young Adults: Target with premium products and convenience-focused messaging to convert their frequent ordering into higher value purchases.
Middle-Income Older Adults: Implement loyalty rewards and health-conscious product recommendations to maximize retention.