Continental’s priority is - among others - continuous improvement in production quality and efficiency.
Data enablement and leverage is key to understanding and managing the process, but it is complex to prepare it for the ongoing process of production.
The data typically comes from several sources (e.g. manufacturing and ERP systems), needs to be joined together, and transfor- med in a way it allows for looking at it from various analytical perspectives. Also, historization is important so that the client can go back into past and see what was the situation at certain given time.
We took the challenge of setting up the cornerstone of Continental’s reporting system properly. The elements of this task included:
➊ Business-related discussion to prioritise the key KPIs (measures) and parameters in which they are looked at (dimensions).
➋ Setting up data structure that reflects corporate and production process hierarchies and dependencies (master data model). Creating parameterised model of the production (lines, products, shifts), business and time periods relevant for reporting - to allow for flexibility and easy maintainability (future changes).
➌ Design data model to store production quality data and its history. Build scripts to create the database entities.
➍ Use of the group-mandated visualization tool Microstrategy (common in the US, but rare in CEE). Design and create metadata objects in Microstrategy to prepare the visualisation environment.
➎ Educate the newly setup analytics team to adopt the best practices for data warehouse and reporting development. Spreading the data awareness into the wider team through training.
➏ Correct data preparation, so they could support a complex process of evaluation of quality in all important aspects (time, line, product, shift, etc). This is an imperative task, because if done incorrectly in the beginning, later changes are very expensive and difficult to do.
Through the series of workshops focused on business and data related discussions we gained a good understanding of the client's production process, generated data and KPIs composition. Also business for quality-management reporting were analyzed. We agreed on the basic list of KPIs, metrics and dimensions.
Then, we led the client team experts through the process of creating the data model along with dependencies and links among entities and tables.
The model was designed with focus on flexibility, maintainability and extensibility. For example, quality information over various time periods can be easily aggregated (e.g. week over week, day in the week over another, working day over non-working day, night shift over morning shift, etc. Or, in case the production line setup changes, the change is only reflected in one place in the database (list of values) and then the rest of data adopts this change and adjusts automatically.
In the next step the team focused on creating the Microstrategy objects. This tool requires to define relationships among data and business entities, filters, etc. While this is time consuming and requires expertise, once done, visualising data into dashboards and reports is relatively easy. Our team had one of the very few real Microstrategy experts within Czech data community.
As an outcome of the project, Client obtained data model covering the selected area in the form that is easily maintainable, extendablenand follows best practices for data warehousing.
At the same time, through hands-on workshops Client’s team was able to co-create these outcomes and learn the process and its rules.
It is yet to be quantified, how the analytics solution will improve the production process. We hope to continuously work with the Continental analytics team to expand the solution and possibly outline the results and improvements in this case-study.
“Cooperation with Colours of Data was highly appreciated by our team. Through a series of workshops, the company supported us in designing of a robust data model of a very complex production KPI. Approach of the company was professional, having expert knowledge in the field and being responsive to our specifications.“
— Kristina Rochová,
Data Analyst, Continental