For two years, Pernod Ricard has been making a data-centric shift by relying on Snowflake and gradually changing public cloud, abandoning AWS in favor of Microsoft Azure.
Claiming second place worldwide in the wine and spirits market, the French group Pernod Ricard has nearly 19,000 employees worldwide and generates 92% of its turnover, approximately 8.5 billion euros, outside France. . It is organized into “brand companies” (six companies producing wines or spirits) and “market companies” (86 companies distributing products, each in a territory). “Two years ago, we started a program to become data centric,” recalled Anaïs Ghelfi, data platform manager at Pernod Ricard during her testimony at Snowflake’s invitation.
Several projects were then launched with specific objectives. Thus, for example, Matrix aimed to optimize marketing investments by using data from both internal and external sources. To deliver the right promotion to the right customer at the right time, Pernod Ricard has implemented Vista Rev’Up. And D-Star was about getting the right product to the right customer at the right time. To carry out this program and these various projects, Pernod Ricard has of course hired data scientists. The group has also set up a data center of excellence (Data CoE, center of excellence) backed by the IT department in order to provide the technical support necessary for each use case identified. This Data CoE has three pillars, the three “Ps”: People (the talents), Platform (the technical platform) and Process (the processes).
One platform for all data
The global technical data processing platform (Global Data Platform, GDP) aims to process internal data connected in real time by API, external data (weather, number of Covid contaminations, etc.) and data imported in the form flat files. This GDP is technically based on the Snowflake solution hosted in the Azure cloud. In all, Snowflake manages 9 TB of data feeding about fifty projects implemented by 200 active users. Finally, a common toolbox was created, essentially composed of Python modules. On the business side, the 7,000 end users have a data exploitation portal based on Microsoft PowerBI on Azure.
“Guaranteeing data quality is very important to reassure users, but it must also be made available and accessible,” insisted Anaïs Ghelfi. This availability should not jeopardize security: we are talking about sensitive commercial data. To properly manage security and access rights, Pernod Ricard has, quite traditionally, defined roles with specific rights, each user being attached to a particular role.
A gradual switch from AWS to Azure
Initially, four years ago, all data was gathered on a Snowflake instance on AWS hosting. This “summary database” is then replicated every twenty-four hours on Microsoft Azure in different databases for the use of each user subsidiary under its responsibility. The choice of Microsoft Azure is justified by the desire to capitalize on group tools (such as PowerBI) mastered by employees.
Gradually, Pernod Ricard is abandoning the instance under AWS in favor of a complete switch to Azure in order to reduce unnecessary costs. Several other developments are already planned. For example, users will benefit from a catalog of data available on their portal. And the Asian continent will benefit from its own instance.