Initially, business questions about cross-country results were very difficult to answer. First of all, because of the mix of business models, procedures and infrastructures differed heavily across countries. The use of different points-of-sale (POS) backend systems forced analysts to use manual exports and analyzing the data required a blackbelt in V-lookup. Furthermore, cultural differences and local offering, made cross-country comparison of products and product categories nearly impossible without a global metadata repository.
To help answer the clients cross-country questions we started a data track. Within this track, our first project was the creation of a global data governance manifest: a comprehensive book of rules and goals, the crystallization of priorities put forward by stakeholders in all levels of the client’s organization.
A data governance manifest is about the value that data delivers, how data gets meaning, becomes consumable on every level within the organization and keeps its uniformity.
A data governance manifest is a first step in making a company a data-driven organization. Therefore, it affects different elements within the company. A data governance manifest provides a shared context for understanding and enriching data, it answers questions concerning data ownership and contains rules & roles that keep the data clean. We started with the definition of different data roles within the company: data citizens, data stewards, a system administrator, a chief data officer (CDO) and a data protection officer (DPO). The manifest describes the specific responsibilities of every role in maintaining the data ecosystem.
We proceeded with the metadata repository, which provides context to the data. It clearly describes all the dimensions, attributes and attribute values used within the data model, visualizes the relationship between the different dimensions and provides naming conventions.
The third and most important part focuses on governance processes. It handles different situations that threatens the uniformity of the data model. These processes handle situations where data is not in line with the business dictionary, technical flows return errors or when the data model needs to be expanded to answer changing business needs. It describes who is responsible and what should be done to maintain the uniformity of the data model.
The second phase within the data track, was the implementation of a data infrastructure and the tools to realize the data governance manifest.
Because every franchise region uses their own flavor and combination of points of sale and hospitality systems, an intensive study and assessment was done of each of the source systems of the regions. This provided us with insights on how the product data was structured in the source systems, how the data could be extracted, etc. Based on these learnings and the data governance manifest a centralized data model was created and hosted in Microsoft Azure.
Based on the regional preferences of the source systems, mapping was done and several Azure Data Factory pipelines were configured to add the data into the central data model on a recurrent basis.
Finally, a data cleansing interface and enrichment interface was created to provide support for the data stewards and the system administrators. This allowed manual actions in an ever-evolving environment with new points-of-sale and new products or the maintenance of important metadata such as opening hours, ingredients, allergens.
In the final phase, a Tableau dashboard for top management was created that visualizes key metrics to evaluate the success of several initiatives taken within the company.