We kicked off the project with a Discovery & UX Design phase. This phase set up the successful delivery of the whole project. We challenged assumptions and evaluated the highest risk aspects of the build through hands on development of narrow proofs-of-concept. This process included reviewing the current use cases and requirements developed by the MNA BI team, digging into source data, and starting to form a view on data models and establishing a data dictionary. The Discovery phase produced a delivery roadmap with revised scope and timelines for all MVPs at a high level with a detailed breakdown of the work for the first MVP. The roadmap also proposed the implementation approach by use case.
This phase was followed up with a two-week Experiment phase. We experimented to provide the perfect testing ground for users to get the first glimpse of what’s possible with the fist use case, a Sales dashboard. In this phase we started to dig further into the technical requirements and set up the required infrastructure. This included the end-to-end flow through of data from the source system, through staging areas, database tables through to the reporting layer.
We took the learnings from the experiment phase and hit the ground running in our first Delivery sprint. We delivered the first minimally viable product (MVP), which was the Sales team’s first report. This report enabled managers and the 100+ sales force to have almost real-time data on hand in the field to make optimal data-driven decisions.
We then delivered the remaining use cases in Finance, Manufacturing, Inventory, and Procurement creating over 20 reports whilst building out the Enterprise Data Warehouse in nine delivery sprints over a six-month period.
The principles that guided our approach to building MNA’s modern cloud data platform were:
- Enterprise scalability, which allowed the data team to grow and adapt with the changing needs of the organisation
- A preference towards performant tools and services, allowing for the processing of both big and small data using the same pipeline
- Security was built-in, drastically limiting the potential for, and impact of data breaches
- Managed services, thus increasing time spent on delivering business value
- DevOps-oriented: tools and services are built around continuous integration/delivery principles, are geared for “fail fail” development approaches and employ best practice agile delivery techniques such as automated testing and self-documentation
- ELT (Extract-Load-Transform) over ETL (Extract-Transform-Load)
- Proven but exciting technology that data teams love to use