Thieme Saves 500 Annual Workdays Using AI and Customer MDM

Thieme utilizes Chat GPT (OpenAI) for automated enhancement of business partner data quality within Stibo Systems’ master data platform. This leads to improved scoring during the process of matching and merging duplicate records and the creation of a well-structured customer database.

“The collaboration with Stibo Systems has been invaluable. Their dedication to understanding our needs and swiftly addressing our inquiries has fostered a truly collaborative and customer-centric relationship. We feel heard and supported every step of the way.”- Dr. Tobias Brockmann, SVP Customer Success & Commercial Operations Thieme

Thieme is a market-leading health care information services provider. Its digital solutions support information, communication and documentation processes in science, education and training, as well as in patient care. Together with a broad network of experts and partners, Thieme is committed to better medicine and a healthier life.

Scattered customer data

The publishing company was facing difficulties in managing customer data spread across seven disparate CRM systems, resulting in duplications and inconsistencies. In response, the company embarked on a mission to centralize customer data into a unified, dependable source. The shift towards digital products underscored the criticality of maintaining clean and structured master data.

Challenges included:

• Very heterogeneous and organically grown system landscape
• Seven CRM systems
• Numerous Point-2-Point direct integrations
• Need to drive digital transformation
• Need for establishing a foundation for data-driven decision making

Challenges

Digital transformation required data consolidation

As a result of many years with organic growth, Thieme had accumulated 400,000 customer records. However, many records were of poor data quality, and there were many duplicates.

Generally, the company was facing a range of challenges related to digital transformation, data management, AI implementation and quality control. To name a few:

Transition from print to digital: While book sales are declining, journal sales remain stable but are also on the decline. This necessitates a transformation of the business model to rely more on digital products, which is a significant challenge for the company.

Integration of business systems : Due to the acquisition of multiple companies over the years, Thieme’s system landscape had become increasingly complex. Seven different customer relationship management (CRM) systems with no proper integration plan or architecture were causing difficulties in managing customer data and conducting analytics.

Data quality and master data management: Matching and consolidating data from different systems, when dealing with variations in company names and other information, was an unwieldy process . Maintaining accurate and consistent data across systems is crucial.

Solution

ChatGPT + Customer MDM = Golden customer records

The overarching goal in Thieme’s digital transformation project was data consolidation and the creation of golden customer records.

To improve data quality, the company integrated an AI-powered language model, ChatGPT, into Stibo Systems Customer MDM to process unstructured customer data and enhance accuracy.

In order to merge duplicate records that might exist in different systems, the company needed to feed the MDM with sufficient data to be able to match the two records.

The integration involved implementing a business action in the MDM to send customer data to OpenAI's ChatGPT API, facilitating data transmission and processing. ChatGPT was utilized to extract and structure key customer attributes for improved matching and deduplication. Through prompt engineering, Thieme can feed unstructured data into ChatGPT, asking for the missing data, e.g., the legal name of the customer, and receive an 85-90% semantically correct output. The improved dataset enables the MDM to match customer data records with 80% accuracy.

With the AI-powered data enhancement and using the match-and-link capability of Stibo Systems MDM, Thieme can create golden records of its 400,000 customers.

The MDM was implemented in coexistence with Thieme’s existing business systems. This means data from, for example, Microsoft Dynamics 365 and SAP CRM are transferred to the MDM for matching and merging and then transferred back. This way, employees can continue working in the system environments they are accustomed to, and with the MDM working in the background, there are no changes in the source systems and the business processes.

Outcomes

Saving time while increasing data quality

Thieme has experienced a highly positive return on investment based on the implementation costs. The speed and accuracy level they have achieved has enhanced compliance and efficiency.

• Improved compliance with GDPR.
• Ensuring consistency across customer and supplier data.
• Improved accuracy and usability of customer data.
• Seamless integration of AI validating up to 96% of data accuracy.
• Substantial time savings through streamlined automation processes, saving approximately 500 working hours annually.

The accuracy of matching records was significantly enhanced, resulting in a boost from 30% to 40% accuracy before the implementation of the MDM solution to 80% accuracy afterwards. This improvement in matching accuracy allows the company to ensure better data quality and consistency across its systems.

Cost savings

The implementation of the MDM solution led to cost savings for the company. With an estimated cost of 8 Cents per record, including infrastructure and service costs, the company could save considerable expenses compared to manual efforts.

The data quality across golden records was significantly enhanced as a result of the MDM solution. This improvement in data quality enables the company to rely on structured data attributes for reporting, analytics and other business processes, making it easier to handle and utilize the data effectively.