Growing a data-culture, naturally, at Roquette

A global leader in innovative plant-based ingredients, Roquette addresses current and future societal challenges by unlocking the potential of nature across the food, nutrition and health markets. Their ingredients respond to essential needs, enable healthier lifestyles and are critical components in life-saving medicines. The company’s approach to innovation extends inwardly and in 2018, they embarked on a digital transformation project to revolutionize their production facilities and processes.

To supply a demanding and rapidly growing market requires efficiency and seamless performance. Pierre-Louis Bescond, Head of Data Science at Roquette, recognized very early on that in a data-heavy industry, transformation would be impossible without the ability to rapidly process and unlock insights from large datasets. With 250 production processes emitting between 500 and 3000 records every 30 seconds, Pierre-Louis implemented Alteryx to consolidate, process and analyze disparate data sources and enable strategic decision-making at scale.

“At Roquette, we are used to processing plant-based raw materials and I see data as my raw materials,” says Pierre-Louis. “With the rise of interconnected systems, sensors and analyzers in all business functions, the ability to derive meaningful insights that drive change requires more than slow, manual data ingestion and analysis.”

Pierre-Louis Bescond
Head of Data Science
Roquette

Having worked in Operational Excellence functions for over ten years, Pierre-Louis is used to seeking out opportunities to enhance efficiencies. “Any team which is forced to spend huge amounts of time manually processing data has the potential to automate using Alteryx,” he explains. “The success of the platform in the Roquette Operations department has created a kind of snowball effect, with more and more people reaching out from teams such as Research & Development, Quality & Controlling, and Human Resources.”

The efficiency gains at Roquette have been dramatic, with one team achieving a 1,000 X improvement on a process which involved the manual export of 3,000 Excel databases. Before Alteryx, the data analysts were using a complicated SQL wizard which needed 15 manual clicks to produce a separate Excel file for each database. With a few iterations and some fine-tuning, Pierre-Louis automated the process within Alteryx and what previously took 100 hours of clicking through spreadsheets is now achieved in just three minutes.

Giving people freedom from what Pierre-Louis calls ‘data slavery’ is the true magic of the platform. Working recently with a frustrated process engineer, Pierre-Louis introduced her to Alteryx to help interrogate the data generated by a newly installed sensor. “She had given up due to the overwhelming amount of operations needed to remove noise and irrelevant data from the source.” Pierre-Louis explains. “But a quick demonstration of filtering in Alteryx provided what she needed to split data depending on product etc. Being able to control your data and quickly uncover areas where you can add unique value, outside of repetitive manual tasks, is empowering.”

The 25 Roquette production sites must be consistently operating at the highest level to reach the exacting specifications of customers. The Sigma Level is a statistical term used in manufacturing to measure how much a process varies from perfection, based on the number of defects per million units. The Excellence team at Roquette was spending large amounts of time gathering disparate data needed to accurately perform the required calculations for the Sigma Level. Pierre-Louis was confident that this time-drain could easily be recovered. “The team needed approximately one working week of data consolidation to assess the Sigma Level of each plant, of which there are 20 in total,” he explains. “With Alteryx, we have been able to replicate this task so that it takes less than three minutes for all plants. This is more than 100 days saved every time we run the workflow!”

Saving time is a major advantage in any industry, but Pierre-Louis also recognizes how machine learning can complement a human role and help uncover vital insights. One team was examining a historical dataset collected from a singular piece of manufacturing equipment. The algorithm indicated that the most important parameter to consider was the external temperature. “Our experts were aware that temperature could have an influence, but the extent of such an unpredictable factor as the weather was unknown and not one the team were able to control.” says Pierre-Louis. “Using the machine learning capabilities of Alteryx, they were able to predict the influence of the temperature on the process and adapt other parameters accordingly.”