Chipotle Leverages Snowflake to Inform Decision-Making During the Pandemic
As reported by VentureBeat, Snowflake connected with Mash Syed from Chipotle Mexican Grill at Transform 2021 to learn how the restaurant is using third-party data for reliable and actionable insights driven by machine learning (ML) during COVID-19. 1
The people at Chipotle believe that food has the power to change the world. With more than 2,850 locations across the United States, Canada, and the United Kingdom, Chipotle’s commitment to “being real” is having a positive impact on its customers, employees, and even the environment.
When the COVID-19 pandemic hit in March 2020, Chipotle wanted to take a preventive rather than reactive approach to restaurant safety procedures using machine learning in order to protect guests and employees. To accomplish this business objective, Chipotle quickly realized the challenge was not building the model but rather having reliable data with near real-time case counts that could be merged with its internal data sets.
Chipotle turned to Snowflake to overcome data reliability issues and streamline its machine learning pipelines—a move that helped the company rank No. 1 in June 2020 for health and safety compliance during COVID-19, according to data from Ipsos’ Consumer Health & Safety Index, as reported by Restaurant Dive.2
Enabling Near Real-Time Access to County-Level COVID-19 Data
Chipotle’s COVID-19 task force needed actionable case count data to inform time-sensitive decisions about safety protocols, supplier constraints, and employee travel. Chipotle’s Lead Data Scientist, Mash Syed, was already exploring a publicly available data source to use for training and building the ML model. Combining COVID-19 data with Chipotle’s internal data made it possible to develop a color-coded, county-by-county risk visualization. The colors represented a classification of whether risk would go up or down in a specific county; green represented lower-risk locations, while orange and red represented higher levels of risk.
Although widely used by other organizations, the COVID-19 data source was unreliable and consistently failed to provide timely insights. Fresh insights were rarely available at the same time each day and, in some cases, not available until the following day. To keep the model results as current as possible, Syed’s team architected a solution that checked for new data every 15 minutes, but data scientists spent more time maintaining pipelines than exploring data. According to Syed, “It wasn’t the machine learning algorithm that was the challenge, it was the data—and, specifically, it was the third-party data source that we relied on.”
Reliable and Actionable Insights that Inform Complex Decisions During COVID-19
Accessing live, ready-to-query COVID-19 data from Snowflake Data Marketplace simplified Chipotle’s data pipeline and reduced administrative effort. “Not only is it a win that data is readily available for stakeholders, but it’s also less of a headache for data scientists,” Syed said.
Reliable COVID-19 insights empowered business users to make data-driven decisions that preserved business continuity and kept customers and employees safe. According to Syed, “We achieved this by putting in place the right protocols, paying attention to COVID-19 case counts, and leveraging data and machine learning to drive decisions.”
Powering Chipotle’s COVID-19 data visualizations with Snowflake freed up technical staff to focus on value-added work. “There hasn’t been a single instance where I’ve had to step in and do extra debugging or maintenance, which is super powerful,” Syed said.
Elevating the Guest Experience with ML
Leveraging Snowflake Data Marketplace, which offers access to more than 500 live and ready-to-query data sets from over 140 third-party data providers and data-service providers, will enable a variety of ML use cases that are next on the roadmap for Chipotle. As an example, combining demographic and internal data will help Chipotle build personalized journeys that impact customer lifetime value (CLV) and recency, frequency, monetary value (RFM). According to Syed, “Curated, reliable data sets from Snowflake Data Marketplace just make things so much easier, and we’re super excited to leverage those data sets to enhance the performance of our machine learning models.”
Learn more about Snowflake’s Data Cloud and Chipotle.
Authors
