Driving organizational efficiency through data unification
"Historically we were not a data-driven organization. Today, with Databricks and OpenValue, we have created a data foundation and are becoming one. That’s the real benefit."
As the spotlight on sustainable engineering and construction grows ever brighter, Colas Group has focused their mission on imagining, building and maintaining transport infrastructure that minimizes environmental impact while maximizing value for their customers. They knew liberating vast volumes of data trapped in hundreds of subsidiaries across 50 countries could contribute to mission success — so they kicked off an aggressive data transformation project. With Databricks and OpenValue, data is now unified, access and quality have been transformed, and best-practice collaboration means data-led innovations become a reality much faster. Insights are creating diverse and exciting benefits, from enabling business as usual during the global pandemic to reducing energy consumption and optimizing asset efficiency.
Hundreds of subsidiaries block access to insights
Colas Group is a highly decentralized organization, comprising hundreds of legal entities around the world. Philippe Toublant is the chief data officer. His team focuses on data architecture, business intelligence and AI projects across the firm. One of the main subjects the team is piloting is known as the “One Colas Data” project and, for Philippe, this meant building a unified data foundation first.
“We were an aggregation of hundreds of companies not sharing data,” explains Philippe, “All creating and gathering gigabytes of data from diverse sources in an array of methods, formats and applications. We couldn’t use our data for the good of Colas Group.” For example, the 54 HR systems meant Colas could neither accurately nor quickly report its total number of employees because the data was so complex to collect and process. And sensor data from operational industrial equipment was created in such volume, it couldn’t be aggregated to create insight and business value.
To address the triple challenge of disparate data sets, data access and data volume, Philippe set out to collate all data from every subsidiary and every source, on-premises and cloud-based. Given the fragmented data landscape, development to production took far too long. Philippe needed a unified platform to deliver insights that create business value much more rapidly.
A unified approach to make data productive
Colas worked with Databricks SI partner, OpenValue, for the project, creating a 100% cloud infrastructure that combined Databricks, Microsoft Azure and various open source technologies. The platform unifies Colas’ data from a variety of data sources, while offering 24/7 availability for operations in multiple time zones. “I experienced how Databricks can drastically increase data productivity,” explains Philippe, “Given our volume of data, Databricks stood out — and the ease of implementation was appreciable.”
Databricks made it possible for Philippe’s team to focus on building business value with data — activities with high added value — instead of spending time on infrastructure and deployment issues. By using the same technologies, the architecture, BI and data science teams are also more closely integrated and collaborate more effectively. Plus, the technology is a catalyst to recruit the best talent. “It’s currently hard to attract young talent. The Databricks and Azure technologies help.”
With the unified data platform in place, the first successes and commercial benefits soon emerged. Philippe’s team helped the Materials and Equipment department reduce energy consumption and increase asset efficiency — an emblematic case with environmental and economic benefits. The team collects data from Colas’ inventory of energy-intensive industrial equipment globally, from all Colas companies and sites, to understand assets owned, usage patterns and energy consumption. By overlaying variables and constraints — like weather conditions, location and legislation — and making the data fully processable within an app, complex usage rates can be calculated to baseline, to monitor and improve asset efficiency. “The global platform means we now have up-to-date visibility of our entire fleet, all equipment and usage rates.”
The collaborative benefits of Databricks were clearly demonstrated during the COVID-19 crisis, a time when remote work was the only option. Interactive notebooks allow the team to work together successfully. “We help each other on tricky code even when it’s no longer physically possible to sit in front of the same computer,” explains Philippe, “We are able to rely on Databricks to keep working in a collaborative way.”
Senior management also needed a solution to monitor the reopening of construction sites after COVID-19–related disruption. Colas’ 15,000 sites began to restart, but data collection was limited due to restrictions on physical site visits. Philippe’s team delivered an app in just two weeks using Databricks Notebooks and Scheduler to automate the data collection process by sending emails and SMS each morning and, based on replies, generate a report on site restart percentages by region. “Collaborative notebooks allow us to share, edit and work together with ease. Without these tools it would have taken far longer.”
Rapid innovation while reducing time to market
Using Databricks, Colas reduced the time to market for its data science initiatives and can now test and deploy use cases rapidly. Some exciting projects in the pipeline include support for procurement — with a dashboard to track global bitumen shipments, allowing Colas to buy product at the correct location, time and price — and support for maintenance with a tool that uses AI to assess video footage of roads and determine the maintenance requirements.
“Right now we’re still experimenting, there are many benefits on the way and more will be realized as time goes by,” says Philippe, who sees Databricks as the enabler. “Before, none of this was even possible. “Historically we were not a data-driven organization. Today, with Databricks and OpenValue, we have created a data foundation and are becoming one. That’s the real benefit.”
