Developing Real-Time Data Products and Platforms
PostNL wanted data products, but each one took over three months to ship. We built the event-sourcing foundation and the programming model around it, bringing that down to two weeks — and new-developer onboarding from weeks to hours.
PostNL
- 2weeks
- to ship a data product, down from more than three months
- hours
- to onboard a new developer, down from weeks
- 3meetings
- an hour each, replacing a non-stop stream of feature requests
About the client
PostNL, formerly TNT N.V., operates mostly in the Netherlands and Belgium. It specializes in the mail, parcels, and e-commerce. The company provides universal delivery in the Netherlands. Additionally, it provides cross-border delivery services. Following the separation of TNT Express from the rest of the company in May 2011, the remainder became PostNL.
As of 2018, the company has more than 37000 employees and earns €2.772 billion per year with €209 million in operating income.
Why were we there?
Trying to compete with its competitors, PostNL realized the value of implementing data-driven logistics strategies. DataChef helped PostNL implement data mesh in response to the limitations of traditional data platforms, training and assisting an internal expert data product team and establishing the standards for data mesh readiness by creating sample data products.
AWS is the main cloud platform provider used in PostNL, and there was no streaming or event-sourcing solution within its infrastructure — though most data sources and applications support webhooks or can communicate with downstream services via SNS or SQS. We designed the event sourcing mechanism around SQS for message delivery, Lambda for processing, DynamoDB for operational access and change events, and S3 for high-retention analytical storage, keeping the migration bar as low as possible.
How we helped
Standardizing data accessibility patterns so the organization has the flexibility to access data based on its analytical and operational needs.
Developing a data product programming model focused on better communication between engineers and the business, letting developers focus on business logic with a strong developer experience, and providing a compass for future data products.
Designing, developing, and shipping data products that attract users from the business and logistics sectors.
The impact we made
Product development time reduced from more than 3 months to 2 weeks.
New monitoring solutions introduced to track cost, codebase quality, and data quality.
A structured intake process replaced non-stop feature requests, with three one-hour meetings supporting delivery instead.
Standardized documentation cut new-developer onboarding time from weeks to hours.
Technologies we used
More case studies
