Fivetran and Snowflake enable business agility for World Fuel Services

Fortune 500 service provider uses Fivetran to pull in data from dozens of SaaS platforms and databases to power business intelligence decisions.

Key results:

  • Real-time business intelligence powered by up-to-date, consumable data
  • 300k new leads generated from consolidated customer data and external sources
  • Visibility on millions of dollars in receivables that were hidden in data silos
  • 200 hours per month saved by eliminating need to manually create, manage and update data connectors

Data Stack:

  • Pipeline: Fivetran
  • Connectors: Salesforce, Jira, Eloqua, SQL server, Postgres, MySQL, Email, and dozens more
  • Destination: Snowflake
  • Cloud Platform: AWS

Data as a Competitive Advantage

World Fuel Services (WFS) is a Fortune 150 company that solves energy challenges by marketing, selling and delivering fuel around the world. Dealing with a volatile market such as energy means that people throughout the organization have to make tough decisions in real time. For example, fuel brokers may have to pull the trigger on a big sale to catch a dip in pricing. The marketing team needs to launch or scale campaigns quickly to take advantage of changes in supply and demand. And delivery routes need to be mapped out precisely to optimize driver schedules. 

Powering Business Insights

WFS’s ability to easily and seamlessly ingest data from dozens of sources with Fivetran empowers stakeholders from across the organization to make data-based decisions that improve customer experience. Here are just two examples:

Global view of customers: WFS has grown enormously over the past 10 years through more than a dozen acquisitions. These subsidiaries operate semi-independently and have their own clients lists that they service — making it difficult to get a global view of customers across the entire company. Now, WFS is able to grab an updated customer list across business units every day to compile a master record that is then enriched by demographic data from Dun & Bradstreet. This master list is then used for lead generation purposes across subsidiaries and business units — leading to upsell and cross-sell opportunities. Because it has this global view of customers, WFS has generated more than 300,000 new leads.

Pandemic response: Travel and fuel consumption took a huge hit during the global pandemic. In an effort to keep revenue flowing into the company, the company made the decision to increase accounts receivable efforts to collect money owed for services already rendered. However, accounts receivable information was spread across dozens of ERP and billing information services across the company’s subsidiaries. Fivetran helped pull data from several disparate sources into a centralized database that provided a centralized view of money owed to the company. The result? The company was able to closely monitor its global receivables every day, bringing visibility on multiple millions in receivables that were not easily accessible before. This was critical in helping WFS navigate the early days of the pandemic.

Before Fivetran: The Need for Real-Time, Scalable Data Analytics

With more than $20 billion in annual revenue, WFS could afford to expend the resources on data integration projects — especially since data was such an important business driver for the organization.

The problem, according to Carlos Mareco, Director of Data Engineering and Business Intelligence, was that the company-built ETL pipelines would pull data in batches, once a day, into an on-premise Oracle database. From there, stakeholders from across the company could query the data using whatever analytics tools they wanted to derive valuable insights. Unfortunately, the Oracle data warehouse grew too large, too fast — making it nearly impossible to run live queries. 

“Additionally, because the data was only loaded once a day, it prevented the real-time analysis that the business needed to stay competitive,” Mareco said. “We had invested in all this data and the infrastructure to pull it together, and we still weren’t able to use it to make real-time decisions. We knew that we needed to change how we consumed data and actually make it usable.”

Snowflake and Fivetran

Around the same time, the company was undergoing a major digital transformation project that would move 22 data centers to the cloud. As part of this effort, Mareco determined that the time was right to scrap the on-premise Oracle database and move to a cloud-based data warehouse.

After investigating several options, Mareco decided to go with Snowflake because of ease of use and the immense elastic scalability of the cloud-based solution.

However, WFS’s legacy ETL solution required Mareco and his team to write custom ETL jobs to bring each table into Snowflake. Even with the vast engineering resources that the company had, this took several hours per table. When multiplied by the hundreds of tables needed, the amount of effort was immense. Mareco was looking at a multi-month project just to pull in all the data needed to run the day-to-day operations of the company. After that, the team would have to continually manage the connectors and update them every time the source made a change to their APIs or database schemas.

“It just wasn’t sustainable,” Mareco said. The business initially used Informatica Cloud to ingest the data because of the team’s familiarity with its on-premise solution, but concluded that they needed something with greater ease of implementation and shorter time to value, and less ongoing maintenance. That solution: Fivetran.

“Snowflake solved many of our data problems by consolidating our data in the cloud in an easily consumable manner, but it didn’t matter if we couldn’t efficiently and quickly ingest the data we needed. It was then that we started to look at different data integration tools.”

Simple, Automated Data Ingestion in Real Time

WFS now uses Fivetran to pull in dozens of data sources from across the cloud — from Salesforce and Box to SQL and Postgres databases. Data is pulled into Snowflake in real time, and users are able to work with whatever data they need to answer critical business decisions.

The simplicity and automation of Fivetran allows WFS to pull in more than 15,000 tables per day — an incredible quantity and quality of information that empowers data-driven decision-making.

“Setting up a connector in Fivetran is easy,” said Mareco. “A department makes a formal business intelligence request that is evaluated on its business impact. If accepted, a Jira ticket is created, and an engineer enters the source’s credentials in Fivetran. The whole process — including approvals, testing, training and rollout — takes just a few days, giving employees the insights they need to make critical decisions in real time.”

Because Fivetran is a SaaS platform, Mareco doesn’t need to update or manage the connectors. So if Salesforce, for example, makes a tweak to its schema, Fivetran schema drift handling handles change management throughout the data pipeline, ensuring that WFS is able to continue to pull in its CRM data to Snowflake without additional configuration — saving the team nearly two days of work for each schema change. Time savings from this alone, Mareco says, is worth its weight in gold since both engineers and users are now able to refocus their time and efforts on other, more strategic projects that impact revenue.

Commencer gratuitement

Rejoignez les milliers d’entreprises qui utilisent Fivetran pour centraliser et transformer leur data.

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.