Next generation of marketing analytics
Next generation of marketing analytics
Has to be reliable and interpretable yet simple.
The market is fed up with one-off solutions and hours of UI clicking to build something worthwhile. Being unstable they require a tremendous amount of effort to evolve.
On contrary, being flexible means you could easily adapt your analytics in ever-changing business environments:
- Express everything as code and version-control it
- Abstract reusable pieces of code into importable modules
- Cover with tests comprehensively
At the same time you want to maintain transparency and predictability to end users:
- Comprehensive documentation on how every number is calculated
- Visual Data Warehouse graph explorer (+ data lineage)
- Metadata access: are we up-to date with sources, which ones bring most of the data
An outstanding solution has to include Software Engineering best practices and core principles: modular structure, code + data testing, continuous integration, self-documenting code.
For these purposes I want to leverage the whole power of cloud services and modern data stack: Azure SQL, dbt, Github Actions, Metabase.
In the next post I will cover the building blocks of my solution.