Insights
Articles, case studies and research on Causal AI in manufacturing: how to find the true causes of defects in production data, and what it delivers in practice.
Articles & Research
AI-Based Root Cause Analysis in Manufacturing: From Mere Suspicion to the True Cause
Statistical process control tells you that something has gone wrong, but not why. This article looks at how automated root cause analysis works, where it reaches its limits, and what object-centric data and Causal AI add.
Read the article ›Survey: AI and Digitalization in Electronics Manufacturing
Xplain Data and SmartRep GmbH · Electronics manufacturing in German-speaking countries
The data is there, its quality is the problem: only 14% of respondents rate their data quality as high. More than 70% plan to invest in AI and digitalization within the next two years.
Learn more
In December 2025, Xplain Data and SmartRep surveyed specialists and managers from electronics manufacturing in German-speaking countries. More than 80% systematically collect quality, inspection and machine data, 64% for more than three years. But only 14% rate their data quality as high, while 71% see gaps or inconsistencies.
Just over a third (35%) already use AI or analytics in production or in pilot projects, mostly for image recognition. The biggest obstacles named are lack of budget (58%), missing interfaces and standards (46%) and lack of know-how (42%). Quality improvement is the top goal (96%). The survey is a qualitative snapshot.
“Electronics manufacturers have moved beyond the phase of simply collecting data. The industry now faces the more demanding task of generating consistent value from it.” (translated from German)Dr. Michael Haft, CEO, Xplain Data GmbH
Case Studies
Siemens: Uncovering Defect Causes in PCB Assembly
Siemens Digital Industries, Erlangen plant · Electronics manufacturing (PCB/SMT)
At its Erlangen plant, Siemens aimed to bring the PCB defect rate below the Six Sigma level. Causal AI revealed design-related defect causes. The investment paid for itself within a year.
Read more
The target was a defect rate below 3.4 per million opportunities. Even though the process was stable, it was not met. Around 500 million data points are recorded every day, for each of the up to 500 components on a board. In a proof of concept, Xplain Data used ObjectAnalytics to bring this data together in one object model: the board as the object, its components as sub-objects.
The causal analysis linked design parameters, solder paste printing results and inspection data. This revealed design-related defect causes that had previously gone undetected. According to Siemens, the investment paid for itself within a year.
“Causal AI is a game changer for us in electronics production.”Erik Schwulera, Head of IoT@Manufacturing and Six Sigma Director, Siemens Digital Industries
Sources: PCEA Magazine, July 2025 · EPP, May 2025 (German)
Schabmüller: Cylinder Head Cover Defect Rate Cut from Over 10% to About 1.6%
Schabmüller Automobiltechnik GmbH, with Schwäbische Werkzeugmaschinen GmbH (SW) · Automotive supplier
Too many cylinder head covers failed the leak test. Causal AI showed that many parts were not leaking at all but had been tested false negative because of the part temperature.
Read more
In the production of cylinder head covers, too many parts failed the leak test and had to go through costly rework. The line has product tracking and is integrated into SCADA and MES. SW, as SCADA partner, and Xplain Data analyzed this data with Causal AI.
The result: the test outcome depended strongly on the part temperature after washing, which in turn depended on the waiting time before the test. Many parts were not leaking at all but had been tested false negative. Quality differences between the casting nests of the raw parts also came to light. After organizational measures and replacing the conspicuous casting nest, the defect rate fell from over 10% to about 1.6%. Hendrik Jacobsen of SW describes the project in an interview with WERKWANDEL magazine (issue 02/2025, p. 30).
“From now on, the application is to be used on all production lines.” (translated from German)Helmut Häckl, Managing Director, Schabmüller Automobiltechnik GmbH
Sources (German): GO>>ing, July 2024 · IT&Production, April 2025 · WERKWANDEL 02/2025, interview p. 30