Technical Showcases - DataLake Projects - Dimedis
Data Lake for Camera Sensors: furniture stores
Company profile
Dimedis is a Germany-based digital solutions partner founded in 1996 with head quarters in Cologne, where around 80 passionate experts leverage years of industry know-how to strengthen and develop business models. Its scalable, “made in Germany” platforms—including Digital Signage, Visitor Management, Event Apps & Lead Capturing, Venue Analytics, and a Retail Platform (Media & Analytics)—integrate seamlessly into customers’ IT infrastructures. Key clients span Messe Düsseldorf, koelnmesse, LANXESS Group, the XXXLutz Group, SPAR Austria Group and many more.
Initial situation
Every day, thousands of customers visit furniture stores across Germany, generating millions of visitor data points that must be processed in near real time. These insights are intended to fuel personalized advertising campaigns for store visitors. The diverse IT landscape—composed of numerous systems and data sources—combined with the high volume and velocity of information, presents significant integration and performance challenges.
Challenges and problems
The main problem was data quality. The data was not processed properly, there were errors due to data duplicates. It was not suitable for further analysis and creation of advertising campaigns. In order to move on to model training, you first need to have clean and correct data. The old solution did not allow for this.
Another problem was the too convoluted and complex IT architecture, which was unreliable and not scalable. For example, the old solution was based on a monolithic database.
Solution
It was decided to implement a data lake—a centralized repository for all raw company data, regardless of its structure or format. To achieve this, a scalable data lake was provisioned on the AWS (Amazon Web Services) cloud platform, enabling seamless consolidation and management of information from the various sensors.
Data Lake Architecture
Additionally, for handling business logic in the final step before dashboard visualization, the low-code platform Megaladata was selected. It enables rapid, intuitive, and transparent adjustments to sensor data processing workflows, making the logic visible across all management levels. Because updates at this stage don’t require constant code edits or rewriting of lambda functions, the solution delivers greater flexibility and reduces the human effort needed to maintain the system.
Some of the AWS components used
Results
The project is still underway and currently in an active development phase. The new system has already been deployed in more than 30 furniture store departments, and rollout to over 120 locations is planned for this year and next.
So far, the data lake is growing by over 500 GB each month. The company generates and stores enormous volumes of data every day, which will serve as the foundation for automated advertising campaign decisions in the future (today, those decisions are made manually based on data from dashboards).
Conclusion
The rollout of the AWS-based Data Lake and the low-code Megaladata platform has helped the company make better use of its large data sets, organize them, and greatly improve their quality. This strong foundation now makes it possible to move on to the next phase—launching automated, machine learning–driven ad campaigns—something that wasn’t possible before because of data errors and low quality.
