Elixr Coffee - Philly, PA
Elixr is one of Philly’s premier coffee roaster
With 4 locations in the metroplex, they’ve found a niche among consumers.
Evan Inatome (owner) is an extremely savvy business owner as well as a coffee sommelier. We spoke early 2018 about working together on a project. He wanted to answer two questions
How can we capture the experience of a first time customer?
How can we optimize pricing?
Of course I had to expand that list
What items are most likely to be paired? (recommendations)
Of items w/ multiple sizes, which are most likely to sell the larger size? (upgrade recommendations)
What items sell most at certain times? (hour/day/season)
Explore customer traffic distribution throughout the day/week
Which items could be removed from the menu?
Data organization- Item misclassification in separate transactions
Hanging out at Center City location
Data Cleaning Process
A project of this nature will always require a monster munging effort. Elixr is a thriving business with over 20,000 transactions per month. These transactions are handled through SquareUp. Downloading the csv files for these transactions are easy. Cleaning this data was a nightmare. Removing artifacts, decifering duplicate names for the same product, etc was difficult. Take a look at the notebook outlining my cleaning and organizing process. The data in this notebook reflects only one quarter of sales for 2019.
Probability Modeling for Promotion Development
Creativity is a major part of data science. I was tasked with finding a way to increase profit for perishable items, expose current customers to new add-on items and encourage new customer engagement. Taking a look at the notebook, you can see taken the quarterly sales and found the probability of of 1 beverage to a particular add on. I discovered the Chai Tea Latte has a high probability of being added with an Espresso Shot. I decided to pitch a week long promotion for a free Espresso Shot with the purchase of a Chai Tea Latte during high traffic hours.
Future Projects - We need to develop a way to evaluate a 1st time customers experience. How do we collect this data and what can we hope to derive from it?