AWS for Industries

Digital supply management using advanced analytics and serverless architecture on AWS

The BMW Group is headquartered in Munich, Germany, where the company oversees 149,000 employees worldwide and manufactures cars and motorcycles in over 30 production sites across 15 countries. This multinational production strategy follows an even more international and extensive supplier network.

Supply chain volatility has significantly increased in the past years: first, driven by the semiconductor crisis, and subsequently by disruptions that affect supply chains on an even broader level. The management of parts shortages and the effective allocation of available parts have become an even more critical and complex task in supply chain and planning teams of automotive companies.

Traditionally, this process is managed by cross-functional committees to work on a joint recommendation. This recommendation provides prioritization of certain car models as well as special options – add-on vehicle features and accessories available to BMW customers. However, the increasing numbers of cases to be managed and frequent adjustments have led to an unsustainable burden on teams. Given that BMW Group offers a large variety of special options across different models and geographic markets, this area has been specifically impacted. Making informed decisions in that space requires data from a multitude of sources, provided by different parts of the organizations, as well as analyses of complex trade-offs across the whole value chain, including, but not limited to supplier and supply management, production planning and financial planning.

To streamline its supply management process, BMW Group joined forces with Boston Consulting Group (BCG) and Amazon Web Services Professional Services (AWS) to develop an analytic solution that integrates supply, production, and financial planning data in one single platform, provides transparency into the entire supply management domain and automatically recommends optimal car parts allocation for special options.

The BMW’s Cloud Data Hub (CDH) on AWS plays a crucial role for the optimizer solution, as data is essential for efficient supply management steering. First, CDH serves as a single source of truth for data. Second, it allows the solution to easily access and combine various data sets to provide required level of transparency and granularity of optimization scenarios.

The solution has been successfully integrated into BMW supply management process and provides cross-functional committees with transparency into the granular volume and financial impact of shortages and allocation scenarios on special option level. Additionally, it facilitates faster, more efficient data-driven recommendations and decisions, unlocking strategic value and direct monetary benefits for the BMW Group.

Business logic

The solution is built in three stages following the existing BMW business process.

  1. Create transparency on the impact of a supply shortage on special options.
  2. Recommend optimal allocation scenarios based on different strategic choices and parameters.
  3. Provide decision support for the resolution of the supply shortage to the relevant decision committees on how to allocate parts or restrict certain option choices.

The diagram below illustrates solution’s business logic.

Figure 1. Digital supply management logic (BCG)

Transparency tool

The foundation for the end-to-end analytics solution is a new level of data integration creating transparency from parts shortage to production volume. The development team brought together information from various data sources found in the BMW Cloud Data Hub (CDH) on parts shortages, bill-of-material data vehicle mappings with weekly volume plans, option take rates, financial data, and market demand. Combining these data assets into one consumable table, allows for transparency on the volume and potential financial impact. This varies from an individual part to special options or packages for each vehicle model that it is configured by an end customer. Resolving this mapping across different levels of granularity and applications was the first major challenge to overcome but also the first value pool that was unlocked.

Optimizer logic

Building on the newly gained transparency, the solution includes, as its second stage, an optimizer module that integrates additional BMW business rules and constraints. On top, it uses further financial information to derive different scenarios of how to best allocate the available parts to models and geographies in the current situation on a special option level. The optimization considers the total demand of a certain part across vehicle models and markets for each individual production week and compares this to the total supply forecast coming from the respective tier-1 suppliers. In a shortage situation, critical models are covered first for strategic reasons or market requirements. The remaining, more flexible demand is then prioritized according to potential financial impact. Within this optimization, different boundary conditions can be set to avoid overly disruptive recommendations, like fair-share logic between geographic markets or models.


The end-to-end supply management process is supported by an interactive frontend application available to BMW supply chain teams allowing inputs of shortage situations, building scenarios around strategy boundary conditions, running the optimization and reporting of the results including financial impacts.

The third stage of the model is the visualization of results using Amazon QuickSight embedded into the frontend application. The solution allows to present results at different level of granularity for different stakeholders, including an overview of results, of required restrictions and possible take rates of certain special options by market and model. Lastly the solution links into a decision support template to help inform the decision committee that decides on production changes.

A crucial feature of the model is its user interaction driven through the frontend: multiple scenarios can be run, stored, shared and new scenarios derived from existing ones. Each scenario is evaluated along business KPIs to select the best suited scenario for a given situation, while allowing experts to steer the model runs according to their experience and qualitative input. The interaction and collaboration are critical to support a business process that spans multiple departments in an end-to-end business process.

Technical solution

The development of the digital supply management solution started by engaging with the line of business and working backwards from end-user needs to design a scalable solution architecture on AWS. To fulfill these needs and to accelerate BMW’s business outcomes, AWS Serverless and AWS Managed services were chosen by the team as the fundamental building blocks of the microservice application architecture.

Figure 2: AWS reference architecture

Figure 2: AWS reference architecture

The AWS reference architecture (Figure 2) demonstrates a secure and serverless solution designed in accordance with the AWS Well-Architected Framework. Due to flexibility and scalability of AWS Serverless and Managed services, the team was able to reduce development effort and accelerate time-to-market of the solution in under 6 months. Moreover, the adoption of these services helped BMW minimize its infrastructure costs and operational overhead in the long-term.

Below, we provide a detailed breakdown of the workflow and the architecture components:

  1. All supply chain domain data required for the solution is available through BMW Group’s Cloud Data Hub (CDH) that serves as the enterprise data lake on AWS enabling data-intensive use-cases.
  2. Upon ingesting multiple data assets, the data passes through pre-processing and aggregation steps of the Data Preparation workflow using AWS Glue.
  3. After data preparation is successfully executed, the processed data is stored on Amazon Simple Storage Service (S3) as a mastertable. It is a single source of truth available to the downstream components including Transparency User Interface (UI) and Optimization UI micro-services via AWS Glue Data Catalog.
  4. From the business user perspective, there are two options to request from the Frontend Application (discussed in 5):
    a. Getting transparency into supply chain data via filtering options. Here the mastertable is queried directly, filtered and cleaned up.
    b. Using an optimization on top of the transparency mechanism to recommend optimal allocation of car parts.
  5. The Frontend Application itself uses AWS Amplify to host a web based application. This application is custom built, but also leverages Amazon QuickSight to embed dashboards with the newly generated data in step 4. The data is accessible via Amazon Athena.
  6. The Transparency UI provides business users with an overarching view of supply, demand and production data in a single dashboard, fostering unprecedented level of transparency and granularity in the supply chain domain. This component consists of an AWS Lambda function that uses awswrangler SDK to query the data from AWS Glue Data Catalog. Amazon DynamoDB table and an error handler logic in AWS Lambda serve as a status mechanism ensuring efficient communication between the Application Dashboard and Transparency UI components through Amazon API Gateway.
  7. After accessing and analyzing the data, BMW planning teams can optimize parts allocation by calling Optimizer UI component. Its core functionality follows a similar pattern as the Transparency UI, with two important differences. Firstly, the data is stored separately using Amazon Simple Storage Service (S3) to aggregate a history of all related optimization runs. Secondly, the core AWS Lambda function uses an open-source optimization library Pyomo that processes the input data, including constraints and boundaries input provided by a user in the frontend application. This function runs a non-binary integer optimization technique to find the optimal allocation points. This setup enables (near) real-time interaction by running Python library in a Docker container hosted in Amazon Elastic Container Registry. Running it in the AWS Lambda function requires between a few seconds and two minutes to finish. That allows users run multiple experiments in a short period of time.

The combination of these components help ensure a seamless user journey as well as a scalable architecture on AWS.

Business outcome

The AWS application provides BMW Group’s supply chain planning teams with a new level of transparency and impact of a parts shortage, consolidating previously fragmented datasets. It provides an automated and highly granular recommendation to decision committees on how a shortage can be managed.

The cross-functional approach with this solution can also be used as a basis for other planning and decision processes. In addition, the developed application simplifies the cross-departmental cooperation. Due to the modular design and micro-service pattern, the application can be easily expanded, serving as a platform for other planning and decision-making use cases.


This blog post describes a supply management solution leveraging advanced analytics on AWS to provide optimized scenarios for parts allocation within shortage situations considering data assets across the entire value chain.

The solution architecture based on AWS Serverless and Managed services helps enable cost-effective operation, interactive visualization, fast optimization engine as well as scalable solution. To learn more about building serverless solutions on AWS visit AWS serverless and Serverless Land.

Michael Wallner

Michael Wallner

Michael Wallner is a Sr. Customer Delivery Architect with AWS Professional Services focused on manufacturing customers and is passionate about the semiconductor industry. On top, he likes thinking big with customers to innovate and invent new ideas with them to transform their business.

Dennis Winter

Dennis Winter

Dennis Winter is a Data Scientist at the BMW Group, with a focus on analytics in supply chain and procurement. He develops cloud-native data analytics solutions.

Dr. Dominik Jäckle

Dr. Dominik Jäckle

Dr. Dominik Jäckle is a Data Scientist at the BMW Group with a strong focus on sales and pricing. His work brings Machine Learning, Human Computer Interaction, and Cloud Computing together.

Dr. Jens Ortmann

Dr. Jens Ortmann

Dr. Jens Ortmann is a Principal Data Scientist at BCG X and part of the leadership team for AI in Automotive. He is an expert for building and industrializing advanced analytics solutions in sales and operations in the automotive sector.

Maik Leuthold

Maik Leuthold

Maik Leuthold is a Project Lead at the BMW Group for advanced analytics in the business field of supply chain and procurement, and leads the digitalization strategy for the semiconductor management.

Shukhrat Khodjaev

Shukhrat Khodjaev

Shukhrat Khodjaev is a Senior Global Engagement Manager at AWS ProServe. He specializes in delivering impactful Big Data and AI/ML solutions that enable AWS customers to maximize their business value through data utilization.

Dr. Tobias Schmidt

Dr. Tobias Schmidt

Dr. Tobias Schmidt is a Partner at BCG and part of the leadership team for AI in Automotive. He is specifically focusing on analytics/AI solutions in supply chain management and volume steering and has supported OEMs and suppliers globally over the past 12 years.