Enhancing Manufacturing with Digital Quality Assurance

Imagine this: An OEM reports a complaint - a safety-relevant component, possibly a faulty screwing process, batch unknown. Your quality manager opens three different systems, searches through Excel spreadsheets and calls production. Two hours later it is clear: the data is there, but not connected. The suspicion: up to 12,000 parts could be affected.

This is not a contrived situation. It is part of everyday life in manufacturing companies that record quality data decentrally - with machine logs here, inspection reports there, batch information in the ERP and plant documentation on paper. Each source is correct on its own. Together, they do not provide a reliable picture.

Digital quality assurance in production solves precisely this problem - not through more control, but through an end-to-end platform that connects all quality-relevant data: Process parameters, test results, component file, worker guidance and archive. This article explains what this means in concrete terms, what requirements such a platform must meet and how to get started.

What digital quality assurance has to achieve in production today

Quality assurance in production has long been a downstream issue: checking what the machine has produced and reworking it if something is wrong. This reaction pattern is no longer sufficient. Stricter product liability regulations, rising recall costs and the growing number of variants require quality assurance that has a preventative effect and provides complete documentation.

Digital quality assurance in production describes the automated, cross-system recording and linking of all quality-relevant process data from the machine control system to the test station and long-term archiving. The aim is no longer control, but transparency: every component has a complete process history that can be called up immediately in the event of an audit, complaint or recall.

What this requires in concrete terms: Data must be recorded with component reference, evaluated in real time and linked to all other systems - MES, ERP, QMS. A reject without a clear cause, an inspection report without a component reference, a screwdriving curve without a batch link. These are all gaps that must not exist in an audit-proof quality assurance system.

Analog vs. digital quality assurance model - a comparison

Why multi-variant production is a particular challenge for quality assurance

An automotive supplier produces 60 different variants of a steering module on one line. Each variant has different tightening torques, different test characteristics and different tolerance limits. The worker has to select the correct instruction for the correct order - from a list of 60 entries. The error rate increases proportionally with the number of variants.

Variant-rich production creates a data volume and coordination problem in quality assurance at the same time. More variants mean more different inspection plans, more tolerance limits, more parts lists and therefore an exponentially increasing probability of errors at the transition points between the system world and the store floor.

Added to this is the shortage of skilled workers: in many manufacturing companies, it can no longer be assumed that every worker knows every variant. The quality assurance platform must take over the transfer of knowledge and ensure that the right process for the right variant is automatically available at the right station without the worker having to search for it.

WHEN A QS PLATFORM IS PARTICULARLY EFFECTIVE

• More than 20 active product variants with different inspection plans on one line
• Less than 80% of worker errors have a clear, documented cause
• Audit preparation takes longer than one working day
• Recall delimitation is not possible without manual research
• More than one system contains quality-relevant data without automatic connection
• Inspection records are not available for specific components

What a digital quality assurance platform must achieve: Core functions

Not all software that records quality data is an end-to-end quality assurance platform. The difference lies in the data flow: a platform automatically connects the levels and ensures that quality data remains available from the production order to long-term archiving without manual transfer.

The following overview shows the four core functions that make up a true QA platform - and why each of them alone is not enough.

Traceability as the foundation of continuous quality control

Traceability is not a feature. It is the prerequisite for all other quality functions to be usable in the event of a fault. A test report without reference to a component is not proof. A screwdriving curve without a batch link is a curve. Only the systematic linking of all data via a common key - batch number or serial number - turns quality data into evidence.

The EU Product Liability Directive 2024 significantly tightens the requirements: manufacturers who are unable to provide complete proof of production in the event of a claim risk a reversal of the burden of proof. This means: No proof = presumed defect. In practice, this means that complete traceability is no longer just an IATF requirement, but a legal minimum.

Typical ROI of a traceability solution in the event of a recall: without complete traceability, a faulty supplier part in the batch of a delivery month leads to a full recall of all parts produced - typically 10,000 to 15,000 units. With precise component traceability, the recall can be narrowed down to the part quantity actually affected. In projects that CSP has supported, the recall quantity has been reduced to less than 5% of the original estimate.

"Our solutions give BMW full control over quality, traceability and processes - for efficient, error-free and future-proof production."

-Korbinian Hermann - CEO, CSP Intelligence GmbH

How CSP works as a digital quality inspection provider in practice

CSP has specialized in quality software for the manufacturing industry for over 30 years. The four modules of the CSP Manufacturing OS - IPM, PG, QST and CHRONOS - can be used individually or operated as an integrated system. The decisive factor is that all modules work on a common master data basis and use the same component key.

According to the company, the suite processes more than 350 billion data points per month in production environments at BMW, Mercedes-Benz, Knorr-Bremse and other customers in the automotive, mechanical engineering, railroad technology and medical technology sectors. The system is manufacturer-neutral: it does not matter which screwdrivers, presses or test equipment are used - CSP connects via OPC-UA, MQTT and XML.

"The introduction of CSP software at Hatz was a great success. The integration of these technologies not only strengthened productivity, but also quality control, which led to an increase in the overall performance of the engine factory."

-Stefan Rotheneichner - Motorenfabrik Hatz GmbH & Co KG ,

Frequently asked questions

What is digital quality assurance in production?

Digital quality assurance in production refers to the use of networked software systems for the automated recording, linking and evaluation of all quality-relevant process data. This includes machine parameters, test results, worker documentation and archive data. The difference to classic quality assurance lies in the continuous data path: each component receives a complete, auditable process history, without manual transfer steps.

Which standards are relevant for digital quality assurance in multi-variant production?

The most important standards are IATF 16949 (sections 7.5 Documented information, 8.5.2 Traceability, 8.6 / 8.6.2 Release decisions) and ISO 9001:2015 (section 6.1 Risk-based thinking, 9.1 Data-driven decisions). In addition, the new EU Product Liability Directive, which introduces the reversal of the burden of proof in claims for damages, has been in force since 2024: Manufacturers without complete proof of production are considered potentially responsible.

What does a lack of traceability really cost in the event of a recall?

Without precise traceability, a partial recall becomes a full recall. For an affected delivery month with 12,000 units produced and an average recall cost of 200 euros per part, costs of around 2.4 million euros are incurred. With component-specific traceability, typically 95% of these parts can be excluded - the actual recall costs are reduced to a fraction. Added to this are class A IATF audit findings, which can trigger an immediate production stop.

How does a QA platform differ from a classic QMS?

A classic QMS (quality management system) documents quality processes, complaints and key figures at company level. A QA platform for production operates at store floor level: it records real-time data directly from machine controls, provides worker guidance depending on the variant and generates component-specific test certificates. Both systems complement each other. The platform supplies the production data, while the QMS manages the higher-level quality processes.

How long does it take to introduce a digital quality assurance platform?

A pilot go-live at an initial production station is usually possible in 6 to 12 weeks, depending on the quality of existing master data and the availability of IT interfaces. Full implementation across all lines typically takes 6 to 18 months. The most common reason for delays is not the software implementation, but the cleansing of inconsistent master data (e.g. different component numbers in ERP and MES).

Can a QA platform be connected to existing ERP and MES systems?

Yes, modern quality assurance platforms communicate via standardized protocols: OPC-UA for the machine level and REST-API for ERP systems such as SAP, Microsoft Dynamics or proALPHA. CSP supports these interfaces natively, enabling integration into existing IT landscapes without having to replace existing infrastructure. The connection to common MES systems is made via the same standard protocols.

What requirements must be met before a QA platform is introduced?

Three requirements are crucial: Firstly, machine connections must be technically possible. Modern screwdrivers, presses and testing systems offer OPC UA interfaces. Secondly, master data (component numbers, inspection plans, variants) must be available in sufficient quality. Thirdly, a clear pilot scope is required: a specific process at a station is the right place to start before the platform is scaled up.