Analytics and AI for SAP Environments: Build a Unified Data Foundation to Drive Advanced Use Cases

ABSTRACT: This blog covers integrating SAP and third-party systems to build a unified data foundation for analytics and AI in conversational use cases.
Sponsored by Boon Solutions
Artificial intelligence enables companies with SAP environments to modernize analytics and foster innovation. This is especially compelling for conversational analytics, which applies generative AI to business intelligence to help analysts boost productivity and managers achieve self-service. But to achieve enterprise-wide insights, companies must create a unified data foundation that integrates SAP and third-party systems. This blog explains how to architect such a foundation, from data sources to integration and governance, then explores the resulting use cases for conversational analytics.
Unified data foundation
We begin by defining the three architectural layers of our unified data foundation: sources, integration, and governance. We’ll work from the bottom up in our diagram.
Architecture
Data sources
SAP functional modules such as CRM, materials management and human capital management serve as the workhorses of the modern enterprise. They automate business processes to streamline operations and enable analytics across cloud and on-premises infrastructure. There is, however, a problem: these SAP modules have proprietary formats, logic and language that make it difficult to extract data from them. This matters because most SAP environments also have third-party systems, including applications, databases (DBs) and content management systems, that hold potential analytics value. To lay the foundation for this, companies must unify tables, documents and so on across SAP and third-party systems.
Companies that unify data across SAP and third-party systems create new analytical insights.
Integrate data
Data engineers integrate their diverse SAP and third-party data in a data warehouse, data lake or better yet a cloud lakehouse such as Databricks or Snowflake. To start, they often ingest all the structured, semi-structured and unstructured data into a landing zone in the lakehouse with an initial batch load. Then they prepare and transform it for analytics. They might filter out unneeded data, enrich certain files, then merge and convert everything to a common format such as CSV or Parquet. They also continuously synchronize these merged stores with their sources, for example by using change data capture technology with source databases. Data engineers, analysts or scientists might merge and format different data stores to support different analytics projects.
Integrating data from SAP sources poses special challenges. For starters, SAP modules have proprietary formats that can be incompatible with third-party tools and APIs. They also have complex logical relationships between tables that can be hard to replicate, which makes the data less readable once extracted from SAP. In addition, the SAP ABAP language is optimized for batch replication but not real-time updates, which makes it hard to synchronize data for time-sensitive analytics. To overcome these issues, data engineers have a few choices. They can build custom transformation scripts, purchase a specialized commercial tool or engage a systems integrator with expertise in such transformations. Only then can data engineers help their companies create a unified data foundation.
Govern
Data engineers and stewards also must govern this diverse data to ensure analytics projects have the trustworthy inputs they need. They validate the accuracy and consistency of database records, append metadata to documents and filter out inaccurate files. They configure and monitor role-based access controls to ensure that only authenticated users perform only authorized actions on validated datasets. They catalog all these assets, track their lineage and document user activities to ensure compliance with internal governance policies and external regulations. All these governance controls depend on well-integrated data that overcomes the SAP challenges described earlier.
Tackle your use cases
Now let’s explore how our unified data foundation enables conversational analytics—and how conversational analytics makes it easier to gain enterprise-wide insights. We consider the experiences of (1) data analysts and data scientists, and (2) business analysts and business managers.
Conversational Analytics Use Cases
1. Data analyst and data scientist
Data analysts and scientists use conversational analytics tools—i.e., BI tools that incorporate GenAI capabilities—to script commands, document data assets and evaluate outputs more productively. This helps them find more insights, faster, across diverse data sources.
- Script. Conversational analytics chatbots help write, test and debug scripts for preparing or querying data. They ask leading questions, suggest options to consider and generate starter code for data analysts and scientists to refine for production usage. This can help a data scientist with a manufacturer, for example, correlate factory performance metrics across SAP and Oracle modules.
- Document. These tools draft descriptions of elements such as data assets, pipeline code and schemas. Data analysts and scientists then review, revise and combine this documentation with other metadata in a central catalog.
- Evaluate. Some tools also use GenAI and predictive ML to help evaluate outputs. For example, a data analyst with an ecommerce firm might prompt her tool to evaluate market trends based on sales comments in CRM records from SAP and Salesforce. She also might gauge sales sentiment to create numerical features for an ML model that predicts quarterly revenue.
2. Business analyst and manager
Business analysts and managers use conversational analytics tools to serve themselves without needing the assistance of data analysts or scientists. These tools generate, recommend and explain analytical outputs, helping the business make faster and more data-driven decisions.
- Generate. Based on high-level user prompts, conversational analytics tools generate analytical outputs. For example, if a factory manager asks which production KPIs changed more than 5% last quarter and why, the tool assesses the contributing factors based on inputs from SAP Advanced Planning and Operation as well as Siemens Insight Hub.
- Recommend. Conversational analytics also recommends metrics, reports or dashboards for users to consider—and even poses additional research questions. If a retail executive wants to add metrics to their weekly sales report, the tool might recommend swapping in more relevant measures from non-SAP systems.
- Explain. While these users understand their business domain, they might need help understanding the data itself. They can have the chatbot explain where data comes from, how to calculate outputs or how different source formats influence the analytical outputs.
Conclusion
By building a unified data foundation that integrates SAP and third-party systems, companies unlock the full potential of analytics and AI. With streamlined data integration and governance, analysts and managers alike gain the insights needed to drive productivity and innovation with conversational analytics. By adopting this approach, companies empower their teams to make smarter, faster, and more data-driven decisions across the enterprise.