Delivering DoorDash Marketing Insights

Introduction

It is hard to believe that it is three years already since the start of the COVID pandemic. Lockdowns and other restrictions meant that people couldn’t dine out and restaurants were looking for ways to stay afloat. As a result, DoorDash and other food delivery services experienced a large increase in demand. Customers who were new to food delivery found out how easy and convenient it was to order online and have the meal of their choice delivered to their location.

In this project, I analyzed DoorDash marketing data to find insights into how customers responded to a recent marketing campaign. The results of this project will help DoorDash optimize their future marketing efforts.

Key Insights

Using Excel, I analyzed a data set of 2,205 customer records. The main insights regarding marketing campaign 6 were:

• 15% of customers responded to the marketing campaign, with an average of $924 spent during the period.
• 71% of the variance in amount spent is explained by customer’s income.
• Customers without children in the home had the highest average total spent at $1,266.
• Customers in the 36-50 age group had the lowest average total spent at $796.
• The average total spent by marital status was not significantly different.

The Dataset

This project is modified from a job interview case study given by iFood, which is the Brazilian equivalent of DoorDash. The data is 98% real but slightly modified for educational purposes. Here is a link to the data.

The dataset consists of 2,205 rows and 36 columns of data. Each row represents an individual customer record. Columns analyzed included:

• Income = customer’s yearly income
• MntTotal = total amount spent by the customer
• Kidhome = number of young kids in the home
• Age = age of customer
• Marital_status = marital status of the customer
• Customer_Days = how many days they have been a customer
• AcceptedCmp6 = did the customer accept offer in 6th marketing campaign
• DateJoined = date customer first became a customer

Analysis

My analysis was done in Excel using techniques such as aggregations, formulas, functions, graphs, and pivot tables.

Since DoorDash was only interested in the results of marketing campaign 6 for purposes of this analysis, I filtered the spreadsheet to only show the records of the customers who had a positive response. This resulted in 333 customer records to analyze, or 15% of the total customers. Aggregation formulas were used to get a summary of the data. A total of $307,828 was spent by customers during the campaign, which is an average of $924 per customer and 1.31% of their yearly income.?

I created a scatterplot to examine the relationship between customers’ yearly income and total amount spent. The scatterplot showed a positive correlation between the two variables, with the total spent increasing as income increases. The R^2 of this relationship is 0.7137, which means that 71% of the variance in amount spent is explained by the customers’ income.?

A pivot table of spending based on the number of children in the home showed that customers with no children spent an average of $1,266 during the campaign. This was 5-6 times higher than homes with 1 or 2 children.

Analysis was also done by age group. Since each customer’s age was included in the dataset, the IF(AND) function was used to create age groups. A pivot table by age group showed that customers in the 36-50 age group spent significantly less with an average of $796. The other age groups spent between $1000-1100. Looking at the age groups split by number of children in the home, half of the customers in the 36-50 age group had children in the home. Because the customers with children in the home spend less, this brought down the average overall for the age group. Interestingly, customers in the age group of 36-50 without children spent an average of $1,342, almost as much as those in the 24-35 age group.

A graph of the average amount spent by age showed similar results. Customers with ages between the mid 20s to early 30s had the highest spending, notably those aged 26 and 32.

Last, I analyzed the data by marital status to see if there were significant differences. The dataset had separate indicator columns for each marital status tracked (divorced, married, single, together, and widow). I created a new marital status column and used nested IF statements so all of the marital status data was in one column. This made creating a pivot table much easier. All of the marital statuses had an average amount spent in the $900-950 range. Breaking this down by number of children in the home showed the same pattern of customers with no children in the home spending significantly more. Customers with a status of Together (living with someone) with no children had the highest average at $1,403.

Recommendations

Of DoorDash’s total 2,205 customers, 333 (15%) had a positive response to the campaign, spending an average of $924 during the period. Further analysis showed areas to focus on for future campaigns:

• Target marketing campaigns to higher income customers. Customer income explained 71% of the variance in total amount spent during market campaign 6.
• Target customers without children. Overall, they spent 5-6 times more than those with children during the campaign.
• Target customers without children in the 24-35 and 36-50 age groups. Both groups spent over $1,300 on average during the campaign.

Additional breakdowns support the recommendation to focus on customers without children. The analysis by age group initially showed customers in the 36-50 range spending the least. However, breaking this down by number of children showed that customers in this group without children spent $1,342 on average. Analysis by marital status showed no significant difference at a high level. Breaking marital status down by number of children was consistent in showing that regardless of status, those customers without children were the highest spenders.

Conclusion

Thank you for reading my article! I really enjoyed using and expanding my Excel skills in this project. This is my first article in my data analysis journey. I would love to hear any feedback you have. Leave a comment below or connect with me on LinkedIn!