Is Your Wallet Shaped by Who You Are?

Unveiling the Powerful Link Between Demographics and Delivery Spending.

Introduction

I was deep in my work when I felt and heard that noise. I wondered if others heard it too. I waited for a beat to see if it would happen again and there it was. My stomach growled and growled again. There was no denying it, I was hungry. I walked to the fridge and let out a groan, I hadn't prepared my food. Now what? Order something, do it. My brain was begging me to take action, and I was in no position to argue. Instinctively, I clicked that familiar red logo on my phone, and within 5 minutes I had food on the way from my favorite sushi place in Seattle. But it wasn't without a price (DoorDash taxes in Seattle are no joke).

And that got me thinking. Did DoorDash know I'd be more inclined to order at this time? Is that why they kept sending me notifications throughout the week? What tips them off about who might be ordering more and when?

I decided to dig into iFood (Brazilian DoorDash) data and I found the answers to my questions and more about how demographics can shape spending. Let's get into it!

What I Learned

After importing iFood's 2018 data into Excel, I cleaned and organized the data to see customers who accepted an offer in the 6th campaign in 2018. Here is what I learned:

1. The total amount spent by customers was $1,139,418 (for this write-up we will use dollars).

2. 67% of the spend variance can be attributed to income.

3. Spending increases with age and reaches a high at age 50.

4. Households with less or no young children spend more compared to those with two young children.

5. Memberships saw the most growth in January and March with 204 and 201 sign-ups, while November and December saw the lowest with 131 and 125.

   
   

The Data

iFood's 2018 dataset which is accessible here, was a use case for an interview process and I was inspired to use it for my data analysis.

This dataset had 40 columns organized by customer demographics and 2022 rows of actual customer information, which were cleaned at the start of the anaysis.

The main columns used in this analysis were:

1. Income

2. MntTotal: Total Amount Spent by Customer

3. AcceptedCmp6: Customers Who Accepted Offer in 6th Campaign

4. Age

5. Kidhome: number of young kids in the home

6. DateJoined: the data the customer first became a customer

   
   

Analysis

After cleaning the data, I organized it by filtering, general aggregations (MIN, MAX, SUB, COUNT), and formulas. The question that came up was who has purchased during this campaign? How might demographics influence spending? With this information, the goal is to provide information to help the company improve its marketing campaign.

I compared annual income with total spending through the scatterplot shown below and used an R-squared value to measure how well we could predict total spend by a customer's income.

We can see that as income goes up, so does spending and it is important to point out the two outliers on the left and right. These would be areas to follow up on to confirm if the data is correct or if these customers are true anomalies.

I also wanted to see how having young children might affect a customer's spending and I used a bar chart to compare households with no children, 1 child, and two children and their total spending.

We can see that households without young children spend significantly at $986,364 than households with young children. We can also see that as the number of young children increases, spend also decreases.

Another question that came up was how age might affect a customer's spending amount so I used a line chart to visualize that data.

We can see that as age increases, spending generally does increase and at the age of 50 is where people spend the most. However, after 50, customer spending starts to trend downward.

Finally, I looked at the time of year that was most successful for membership signups.

The data shows that January and March brought in the most signups while November and December saw the least. This data can help us ask better customer questions about the why. Was it because people were tired after the holidays that the beginning of the month was the most popular? Do the holidays influence the lower signups in November and December?

Conclusion

I set out to analyze iFood's marketing data to understand what business opportunities and insights we could get from campaign 6 customers. Based on my analysis, the company's marketing team should target higher-income customers, especially during January and March. They may also want to consider how they're appealing to certain age groups, especially those aged 50 and customers without young children in the home.

This was my first data write-up with Data Career Jumpstart and I had a lot of fun working with Excel to analyze the data.

Now I'd love to hear from you! What would you have wanted to know more about? What feedback do you have?

Feel free to comment below or connect with me on Linkedin.

This project was done as part of the DAA Boot Camp Projects. Educational purpose.