Mosaique 1.0
The problem
Complex data processes slow down marketing teams
Every day, marketing teams are faced with the challenge of making data-driven decisions quickly and efficiently. Reality, however, confronts them with complex and resource-intensive data integration into declining systems. They must complete comprehensive processes from extracting data to loading, integrating and transforming to modeling and reverse ETL — tasks that require a deep understanding of data engineering and data science.
In numerous customer projects, we have found that these technical requirements represent a significant challenge. Many teams do not have the necessary resources and specialized marketing and CRM know-how, either internally or within their overall organization, to effectively master these tasks. This often leads to delays, high investments and sub-optimal results, as important details in the requirement description of the use cases are overlooked or underestimated.
Our solution: Mosaique 1.0
To address these challenges, we have developed Mosaique 1.0. Our solution and architecture is designed to take on the complex and technically demanding steps of data processing, from extraction to reverse ETL. Based on an order stream, Mosaique 1.0 calculates precise customer profiles, customer lifetime value (CLV) and dynamic segments. The model not only simplifies the day-to-day work of marketing teams, but also enables them to focus on what they do best: creative and effective marketing based on data-driven insights. In addition, our solution is designed to be headless and can be integrated into all common visual and BI interfaces with just a few clicks, which enables seamless and flexible use in existing systems.
Mosaique 1.0 capabilities
1. Data extraction and anonymization: Mosaique 1.0 starts with setting up on an order stream, which we can extract from any interface. Users themselves define the desired period of time, such as the last 12, 24 or 36 months, which we then build on on a rolling basis. During the charging process, we transform and anonymize all personal data. Order IDs are encrypted and so are customer numbers or email addresses to ensure data protection, security and flexibility of the model from the start.
2. Safe data storage: Once loaded, the order history is completely anonymized and securely stored in a PostgreSQL database in the Google Cloud with server location in Germany. This is done in compliance with the highest safety standards and protection mechanisms. Marketing teams don't have to worry about operating and managing servers, Docker containers, auto-scaling, or Kubernetes — we take care of these technical aspects.
3. Data processing and customer analysis: Customer numbers or email addresses that have already been transformed into UUIDs are grouped and combined to form an encrypted customer profile. Based on this grouping, we calculate the Customer Lifetime Value (CLV), the order frequency and the timeliness of the customer relationship for each UUID, measured on the days between the last order and the present day.
4. Scoring and segmentation: Profiles are then rated based on CLV, order frequency, and timeliness, with three scores assigned to each profile. The scores range from 1 to 5, with 1 describing a low CLV, high frequency and timeliness of 5. For classification, we use quantiles to classify profiles into segments compared to all others.
5. Flexible customer segmentation: In the last step, our model offers flexible and dynamic customer segments. By default, there are nine segments that categorize customers according to their CLV, their order frequency and the timeliness of the customer relationship. This segmentation enables marketing teams to implement targeted strategies effectively and efficiently.
The output of Mosaique 1.0: Detailed customer segments with our API
Mosaique 1.0 uses its API endpoint to generate a structured JSON object that contains a list of arrays. Each array represents a customer segment with detailed statistical data. This allows deep insight into various customer categories, which are decisive for targeted marketing strategies. Here is an example of the output structure:
● Group: The name of the segment (e.g. “new customers,” “VIPs,” “shaky candidates”), which describes the type of customer segment.
● Number of UUIDs: The number of customer UUIDs in this segment.
● Order Count: The average number of orders per customer within the segment.
● CLV (Customer Lifetime Value): The average lifetime value of a customer in this segment.
● Timeliness: The average number of days since customers last placed an order in this segment.
● Percentage: The percentage share of this segment in the total customer base.
By using the API output of this detailed data object, marketing teams can seamlessly integrate the information with any BI and visualization tool, such as Tableau, Google Sheets, or Microsoft Power BI. This makes it easier to dynamically visualize and thoroughly analyze customer data, which in turn supports strategic decisions and targeted marketing measures.
Additional support and tailored solutions:
We've often seen that not every team has the resources or technical ability to manage data integrations on their own. That's why we not only offer a powerful API, but also comprehensive support in visualizing the data. Since visualizing data ourselves is very important to us, we have developed our own dashboard that provides an intuitive and accessible presentation of the segments.
For teams who want to use our solution directly without integrating it into existing BI systems, we are ready to help create and adapt visual interfaces. In addition, we offer support in integrating our API into common BI systems. This includes both technical assistance and advice to ensure that you get the most out of Mosaique 1.0, no matter what environment you're working in. Through these flexible support options, we guarantee that every marketing team, regardless of technical equipment or previous knowledge, can take full advantage of our advanced data analysis.
