Drinking Data

Today Big Data is everywhere… even in our drinks. Senseable City Lab took on a challenge by the Coca-Cola Company: can we find any value in the huge amount of information recorded by their Freestyle dispensers? There are more than 15,000 Coca-Cola Freestyle units in the USA, each with the possibility to serve over 150 unique drinks. Each transaction is recorded as a “data string” – including time, location, and user preference. Our task was to put our hands in it...We began by visualizing patterns of total consumption, where we found spikes during weekends and dips during weekdays – although people are less likely to fill up a cup on special days like Easter and 4th of July. Brand loyalty and proportions are also striking, although they tend to vary across the United States.

But even more compelling is the translation of big data on the ground - from terabytes of digital strings to restaurants, cafeterias and truck stops. Today we can use big data exploration to better understand what happens in the real world: zooming in to individual dispensers shows unexpected traits. For instance, people are loyal to specific brands… but only if someone is watching. The type of service (self or crew) has a huge impact on patterns; when users have direct access to a self-service dispenser, they tend to explore, mix and match, but when a waiter is filling up, people are much more shy. Also, several machines with the highest consumption in the USA serve a disproportionate amount of water – owing to their restaurants’ policy of sending a carafe out to each table.

Drinking Data – just a sip of what could indeed turn into a substantive framework for translating data analysis and visualization into physical insight – and back, as people respond. Come back to check out our next projects on senseable cities as part of our ongoing collaboration with Coca Cola.