In a data use case workshop, taod develops a proof of concept together with Blinkist's finance team. More than half of the budget is invested in marketing or represents a necessary investment for effective marketing. Ultimately, therefore, a clear allocation of investments must be made so that it is clear where exactly what money is being spent. Existing discrepancies between self-generated reporting and actual accounting must be eliminated. This use case should form the basis for further optimization of the analysis activities. 

During the workshop, it quickly becomes clear that Blinkist's data stack needs to be fundamentally overhauled. A high susceptibility to errors was identified in the existing data pipelines, which had to be completely rebuilt. Changing the connector tool from Data Virtuality to Fivetran via Airbyte on a trial basis, primarily to relieve the burden on internal resources, and the transformation tool from Matillion to dbt will ensure enormous flexibility in future and allow any adjustments to be made quickly and easily. 

As a transformation tool, dbt reduces the effort required for thorough testing by up to 80 percent. In addition to the complete overhaul of the data transformation and the change of connector tool, taod designs new logics to meet the dedicated analysis requirements. The data vault method is used. It ensures the fast and correct integration of data into a data warehouse. It can react flexibly to major changes in content.

The existing Amazon Redshift data warehouse will be replaced by the Snowflake cloud data warehouse, which will ensure high scalability as required in future. The development of the data stack will initially be managed and subsequently supported by taod, thus ensuring a high level of enablement for Blinkist. Data Vault also represents a new modeling technique for Blinkist's data team, which will be trained through 1:1 coaching sessions over the course of the project. Technology and enablement also reduce the time needed for bug fixing. Whereas corrections used to take days at worst, bugs can now be identified and fixed within minutes.

After a project duration of around four months, Blinkist's data engineers, data scientists and data analysts manage the new tech stack independently. From this point on, taod assumes an advisory role and takes care of the continuation of the project in terms of content at regular intervals as part of an FAQ format. The updated Modern Data Stack ensures the smooth technical processing of the complex data. The high susceptibility to errors has been reduced to a minimum, partly because the new modeling technology in combination with modern ETL processes significantly reduces sources of error from the outset.