HelloFresh’s data journey: Insights from scaling to 1 billion meals (SEG207)
How HelloFresh Built a Data Platform to Power Growth and Data-Driven Products
Introduction to HelloFresh
HelloFresh is a globally integrated food solutions group, operating in 18 markets across three continents.
The company was founded in 2011 as a meal kit business in Berlin, with a mission to change the way people eat forever.
HelloFresh operates a brand portfolio with nine different brands, covering various customer needs and verticals, such as home cooking, convenience, and pet food.
The company has experienced accelerated growth in recent years, with the COVID-19 pandemic driving a significant surge in demand.
Challenges with Rapid Growth
As HelloFresh grew rapidly, the company faced challenges with handling the increasing volume of data and evolving business contexts.
The centralized data team became a bottleneck, unable to keep up with the pace of change and the growing data needs across the organization.
The initial attempt to implement a data mesh approach did not yield the desired results, leading to a "data mess" rather than a data mesh.
Lessons Learned and Iterating on the Data Platform
The initial data mesh implementation failed due to a lack of a clear playbook, investment challenges, ownership and domain understanding issues, and a "lift and shift" approach without considering a global data model.
The team realized the need to bring different personas within the organization (backend engineers, product managers, etc.) along in the data platform journey, as not everyone had access to data experts.
The Tardis Data Platform
HelloFresh built a multi-model, unified data platform called Tardis to address the challenges.
Key features of Tardis:
Seamless data integration and ingestion from various sources, including event-driven data from Kafka.
A global data model that represents the business's operations.
Simplified onboarding and abstraction of complex tasks (infrastructure, permissions, etc.) for users.
Cost optimization capabilities to leverage the best AWS services.
Automated data modeling and transformation capabilities.
Emphasis on training and documentation for platform adoption.
Use Cases and Benefits
Tardis enabled the seamless ingestion of event-driven data from Kafka, reducing the time to bring data into the analytical plane from months to days.
The platform provided a simple interface for backend engineers and data analysts to create data products and access the global data model, reducing the cognitive load and dependency on specialized teams.
Tardis supported the implementation of a streaming data capability, allowing for near-real-time data processing and analysis.
The platform adoption grew by over 300% within the organization, with more than 650% growth in the number of data assets created.
The platform enabled use cases like a cost analysis dashboard for annual planning and a personalized loyalty program for customers.
Key Learnings
Decentralization should be implemented with a clear purpose, not just as a buzzword, and involve all relevant stakeholders.
Measure success based on platform adoption, not just the number of features or data assets.
Focus on real-life use cases and work backward from the needs of internal and external customers to ensure positive ROI.
These cookies are used to collect information about how you interact with this website and allow us to remember you. We use this information to improve and customize your browsing experience, as well as for analytics.
If you decline, your information won’t be tracked when you visit this website. A single cookie will be used in your browser to remember your preference.