Instacart Market Strategy Analysis
Pandas | Matplotlib | SciPy | Seaborn
Project Overview
-
My task is to derive insights sales data for the grocery app, Instacart, in order to better segment customers profiles and suggest marketing strategies. Key business questions include:
How should ads be scheduled to avoid busy times? What are high sales times/days?
How should price ranges be simplified across products?
Which products are most popular?
What does customer behavior look like across brand loyalty, ordering habits, region, age, family status, and income?
-
Includes name, gender, state, age, number of dependents, family status & income
-
Data wrangling & subsetting with Python (pandas, data structures)
Combining and exporting (.merge, .join)
Deriving, grouping & aggregating variables (if statements, for-loops, user-defined functions, agg(), loc(), transform())
Data visualization with Python (matplotlib, scipy, seaborn)
Coding etiquette
Analysis
What does behavior look like across age, income & class?
Age does not influence the number of dependants but it is a predictor of income. People over 40 are much more likely to make over $200K & people below 40 make mostly below $200K. Further, class has a marginal impact on average price paid for certain departments. Lower classes pay a lower price for meat & seafood, international, pets & beverages and a higher for price the items in the "other" department.
Analysis
Are age and family status correlated?
Generally, price increases as age increases for divorced and widowed people but the trend is absent for single people.
Those living with parents and siblings see a decrease in price as they age. Likely the result of becoming financially independent as they age into adulthood.
Recommendations
How should ads be scheduled to avoid busy times? What are high sales times/days?
Instacart should schedule ads from 1 - 6am, Mon - Wed to avoid high order volume periods like weekends and work hours. Order prices are highest from 12 - 4am and on weekends. Instacart should advertise alcohol, indulgent snacks and large price / quantity items (party-size bags & cases of beer) given the increased willingness-to-pay at this time.
How should we categorize prices for simpler analysis?
Instacart should use 4 buckets defined by the statistical quartiles of the price metric.
Which products are most popular?
Produce then eggs & dairy are the most popular. We should use department to further segment customers (ex. produce is a grocery order & alcohol is a convenience order) and market to each customer type with unique messaging.
Are there differences in ordering habits based on a customer’s loyalty status?
There is no significant differences in neither mean price nor department id based on a customer's loyalty status. If this trend maintains for all behaviors, we shouldn't use loyalty to segment customers.
