“With Fivetran and Snowflake, Sennder is not only a real data-driven company, but we’re fully prepared for another hyperscaling phase.” - Christian Haas-Frangi, Staff Software Engineering Manager, Sennder
- Saved at least €100,000 per year, enabling its team of engineers to focus on data modeling and analysis rather than pipeline maintenance
- Enabled more than 80 percent of Sennder employees to drive decisions using the 30,000+ dashboards powered by Fivetran
- Gained reliable access to data to train machine learning models that will help predict costs for shippers or personalize offers to carriers
- Pipeline: Fivetran
- Sources: 200+
- Destination: Snowflake
- BI Tool: Looker
- Orchestration: Airflow
- Data Transformation Tool: dbt
Today, there are more than 6.2 million trucks in circulation through the European Union, and the logistics market largely still depends on paper, phone and fax. Sennder, Europe’s leading freight forwarder, is on a mission to revolutionize the industry by leveraging data and technology.
With sennder, shippers benefit from simple and efficient access to an otherwise scattered fleet of 40,000 trucks to tender their freight via electronic data interchange (EDI), API or in-house developed automated broker services. Carriers search and book loads via sennder technology, while drivers use an app to ensure full capacity and arrive at the target destination on time. These offerings are enabled by data-driven apps, machine learning models and BI dashboards.
Building a modern data stack with Fivetran and Snowflake
Since the company was founded in 2016, sennder had an established central data team of two to three data analysts to manage all analytics projects. In 2019, two data engineers were hired to build a data warehouse with Snowflake, and the team also brought on Fivetran to automate the ingestion of data from its ERP system and thousands of Google Sheets.
Soon after, sennder’s adoption of Fivetran grew to additional use cases, including improving the efficiency of invoice generation. Generating an invoice was a very manual process because data was managed by a backend tool that required its engineers to manually build a data pipeline. However, with Fivetran’s out-of-the-box connectors, sennder was able to easily ingest and centralize the data from the backend tool into Snowflake in a matter of days. Sennder’s data engineering team also leveraged Fivetran to refocus on higher-value projects. At sennder, ad-hoc reports are self-service and generated by users directly in Looker or, if something more complex is required, by data analysts.
“Fivetran freed our small team of data engineers from building and maintaining data pipelines, enabling them to focus on higher value work like developing complex and data-driven products such as customer-facing analytics platforms or recommendation systems.” - Christian Haas-Frangi, Staff Software Engineering Manager at Sennder
As the company experienced hypergrowth, increasing from 50 to 500 employees in less than a year, the amount of data being collected increased at a tremendous rate. Business logic increased in complexity and the need for reporting grew. It was impossible for one team to handle all data engineering projects.
Sennder met this challenge by dividing the central data team: One data platform team that handles the core aspects of the data stack and an analytics team responsible for each business department (e. g. shippers, carriers, finance and marketing).
Moving to a data mesh
When 30 data analysts contribute to one big repository that contains around 2,000 data models — there’s always the risk that a new creation interrupts operations.That’s why sennder moved to a data mesh approach. The sennder data lake contains raw data, which is updated as frequently as possible. The data engineers and data analysts configure the synchronization in Fivetran within minutes.
The clean data is stored in around 30 repositories, called data marts, which are part of the Snowflake Data Cloud. All data marts are connected and owned by the analytics team of the business area where the data was generated. A data mart contains the respective data models, plus computation space, so each analytical team can run its transformation and modeling without affecting the other teams.
All the data ingestion, transformations and modeling run automatically. Thanks to the newly released Terraform module in Fivetran, they’ll soon forget having to manually define data marts as a destination in their code.
Democratized data access drives innovation
Today, everyone at sennder has access to data for their daily jobs. Christian’s team has enabled the creation of thousands of self-service BI dashboards to serve leadership and all lines of business across the company. Engineers and data scientists also have access to fresh, reliable and secure data to train machine learning models to predict costs for shippers or to personalize offers to carriers.
Furthermore, democratized access to data is powering sennder’s Ship Green initiative which aims to reduce CO2 emissions by offering shippers a tool to determine their CO2 emissions and define measures to reduce them.