Image credit: Sensibill

Have you ever wondered which one of your coworkers shares your taste in takeout? Or how about your controversial love of pineapple on pizza?

Well, our team has, so we decided to look into it—using our own spend data, of course. 

As a fun, Valentine’s-themed activity that allowed us to pair our knowledge of data and spend with a desire to answer the age-old question, “can two people truly agree on the same toppings for pizza?” we dove into our own spend data to figure out who our Sensibill food soulmates are. 

Here’s how it worked

Sensibillians submitted a minimum of three restaurant, food, and takeout receipts for our team to analyze.

Our talented machine learning team took our receipts, conducting experiments, and applying analytical methods to find each team member’s food "soulmate". 

The data helped us narrow down “soulmates” into categories of teammates we should and should probably not try and split a pizza with. By using word embedding to identify similar items (like burgers and fries), we created pairings or “matches” between teammates based on highly similar purchase-level spend. 

The results for both categories were based on similar and dissimilar items, deduced through analyzing purchase-level data.

Here’s why we did it

There’s a lot that purchase-level data can reveal about the way people spend their money. It can, for example, detect patterns in people’s spend and illuminate different spend habits and purchase behaviors. 

In our case, we wanted to apply our own knowledge of purchase-level data, along with the capabilities of our products, in a fun and unique way to experiment with different use cases for purchase-level data. 

It’s only by looking at deeper, more contextual data like this that you can truly understand how and why people spend the way they do. 

Especially when you’re trying to understand just why the heck two people can’t agree on pizza toppings!

Here are some fun insights

When analyzing the data from our team’s purchase-level spend, we discovered a few interesting insights. 

In particular, we learned that context is crucial. For example, just because team members ordered rice from many restaurants (such as Indian and Chinese establishments) doesn’t mean two team members have the same tastes in food. This meant looking more closely at purchase-level spend that showed multiple similar items (versus just one or two!). 

Oh, we also learned: 

• One pairing shared both a love for the Rude Dude burger and brownies

• Another pairing was matched for a fondness of poutine and burgers 

• A different pairing was matched based on consistently ordering Na'an and other Indian dishes

• For our anti-matches, one team member particularly preferred Italian, which made them a less ideal restaurant match with a sushi-obsessed teammate!

One thing became abundantly clear throughout our analysis—at Sensibill, we love burgers!

 So, the next time you’re wondering which coworker to grab lunch with, ask them for their receipts and check out their purchase-level spend. You may just find your food soulmate ✨