For years, quality management has asked organizations a relatively simple question:
What could go wrong?
The next generation of quality management is beginning to ask a more ambitious question:
Can we identify the warning signs before something goes wrong?
This represents an important shift in the way organizations think about quality.
Risk-based thinking became much more prominent with ISO 9001:2015. The standard moved organizations away from relying primarily on corrective action after problems occurred and toward identifying risks and opportunities that could affect the intended results of the quality management system. ISO describes risk-based thinking as an essential element of an effective QMS.
Now, with ISO 9001:2026, the direction has become even clearer.
The new edition retains risk-based thinking but provides a clearer distinction between risks and opportunities, encouraging organizations to proactively address both as part of decision-making and continual improvement.
This raises an important question for quality professionals:
Is the future of quality management simply better risk management — or is it predictive quality management?
From reactive quality to predictive quality
Traditional quality management can be viewed as a sequence:
Problem → Inspection → Nonconformity → Corrective Action → Improvement
This model remains important.
If a concrete cube fails a compressive-strength test, for example, the project team investigates the cause, determines corrective action and prevents recurrence.
But by the time the cube has failed, the undesirable outcome has already happened.
Risk-based thinking changes the sequence:
Risk identification → Risk assessment → Preventive action → Monitoring
This is more proactive.
But predictive quality goes one step further:
Data → Early warning → Prediction → Intervention → Quality outcome
The difference may appear subtle, but it changes the role of the quality function.
Instead of asking only:
“What risks do we have?”
the organization begins asking:
“What measurable signals indicate that a quality problem is becoming more likely?”
That is the beginning of predictive quality management.
What exactly is predictive quality management?
Predictive quality management is an approach in which historical and real-time information is used to identify patterns, trends or warning signals that may indicate a future quality problem.
It does not necessarily require artificial intelligence.
An organization can begin with relatively simple techniques:
trend analysis;
control charts;
Pareto analysis;
regression analysis;
statistical process control;
leading indicators;
supplier performance trends;
inspection data;
complaint patterns;
defect recurrence analysis.
More advanced organizations can use:
machine learning;
anomaly detection;
predictive analytics;
digital twins;
automated inspection;
sensor data;
AI-assisted quality analysis.
The important principle is not the technology itself.
It is the transition from detecting failure to detecting the conditions that precede failure.
The limitation of the traditional risk register
Many organizations already have a risk register.
It may contain entries such as:
| Risk | Probability | Impact | Mitigation |
|---|---|---|---|
| Concrete strength failure | Medium | High | Increase testing |
| Supplier delay | Medium | High | Monitor supplier |
| Waterproofing defects | Medium | Medium | Increase inspection |
| Design error | Low | High | Design review |
This is useful.
But there is a potential weakness.
A risk register can become a static document.
The organization identifies the risk at the beginning of the project, assigns a probability and impact, identifies mitigation, and then reviews the register periodically.
The problem is that risk does not remain static.
A supplier that had a 5% defect rate in January may have a 12% defect rate by June.
A subcontractor may begin accumulating inspection failures.
Concrete temperature may progressively increase during hot weather.
Rework may increase from 1% to 3% to 5%.
Customer complaints may begin clustering around one product.
These changes may represent signals that the probability of a future quality problem is increasing.
A traditional risk register may not capture this dynamic movement quickly enough.
From risk register to risk intelligence
This is where predictive quality management becomes interesting.
Instead of treating risk as a static entry, organizations can connect risk to measurable indicators.
For example:
Traditional approach
Risk: Waterproofing failure
Probability: Medium
Impact: High
Mitigation: Increase inspection
Predictive approach
Monitor:
number of waterproofing defects;
defect recurrence;
installer productivity;
substrate moisture;
inspection rejection rate;
weather conditions;
material batch information;
previous repair locations.
The organization can then identify whether these indicators are moving toward an undesirable pattern.
For example:
Inspection rejection rate
2% → 3% → 4% → 7% → 9%
The organization should not wait until widespread waterproofing failure occurs.
The trend itself becomes a warning signal.
This is the difference between risk documentation and risk intelligence.
Leading indicators become more important
One of the biggest changes associated with predictive quality is the growing importance of leading indicators.
A lagging indicator tells us what has already happened.
Examples include:
number of defects;
number of complaints;
rejected materials;
failed tests;
rework cost;
NCRs;
warranty claims.
A leading indicator attempts to show what may happen next.
Examples include:
inspection rejection trends;
percentage of overdue inspections;
subcontractor competency;
calibration status;
percentage of unresolved technical queries;
frequency of repeated defects;
supplier delivery variability;
percentage of work performed without approved inspection;
design changes occurring late in the project.
The distinction is important.
A project could have:
Low NCRs today
but simultaneously have:
Increasing inspection rejection rates.
The first indicator looks positive.
The second may provide an early warning.
A mature QMS therefore needs to look at both.
Construction provides a powerful example
Consider a large building project.
The project has a KPI called:
Concrete Cube Test Pass Rate
Suppose the results are:
| Month | Pass Rate |
|---|---|
| January | 100% |
| February | 100% |
| March | 98% |
| April | 97% |
| May | 94% |
A traditional approach might react when the pass rate falls below the acceptance criterion.
A predictive approach asks:
What caused the trend to deteriorate?
The organization could investigate:
concrete temperature;
batching plant performance;
cement source;
aggregate moisture;
mix design changes;
transportation time;
placement time;
curing conditions;
testing laboratory performance;
subcontractor practices.
The objective is not simply to record that the pass rate decreased.
The objective is to identify the conditions associated with deterioration.
That information can then be used to intervene before the problem becomes significantly larger.
AI can accelerate this process
Artificial intelligence can potentially make predictive quality management much more powerful.
Imagine a project with thousands of quality records.
The data could include:
inspection requests;
material approvals;
laboratory results;
NCRs;
photographs;
weather information;
subcontractor performance;
equipment records;
progress data;
design changes;
RFI records;
audit findings.
A human quality engineer may find it difficult to identify relationships across all of these datasets.
AI and advanced analytics can help identify patterns that deserve human attention.
For example:
“Projects using Supplier A under conditions X, Y and Z have historically shown a higher frequency of defects.”
Or:
“The current combination of inspection rejection rate, subcontractor performance and rework frequency resembles patterns observed before previous quality deterioration.”
The system is not necessarily declaring that a failure will happen.
Instead, it can provide an early warning.
The final decision remains with the quality and project management professionals.
Predictive does not mean certain
This distinction is extremely important.
Predictive quality management does not mean that an organization can know the future with certainty.
A prediction is based on information and patterns.
It can be wrong.
For example, an AI system might identify a high probability of a defect because similar historical conditions previously produced defects.
But the current project may have additional controls that prevent the problem.
Therefore:
Prediction ≠ Fact
and:
AI warning ≠ Nonconformity
A predictive QMS should treat predictions as decision-support information, not as unquestionable conclusions.
Human review remains essential.
ISO 9001:2026 does not require AI
It is important not to misunderstand the new direction.
ISO 9001:2026 does not mean that every organization must implement artificial intelligence or predictive analytics.
The standard remains a management-system standard.
ISO 9001:2026 continues to emphasize customer focus, the process approach, risk-based thinking and continual improvement.
The new edition provides a clearer distinction between risks and opportunities and encourages organizations to take proactive action.
Therefore, an organization can develop predictive quality capabilities without claiming that AI is an ISO requirement.
The technology is optional.
The management principle is more fundamental:
Use information to make better quality decisions before undesirable outcomes occur.
ISO 31000 and the wider risk-management ecosystem
Predictive quality management also fits into the broader risk-management discipline.
ISO 31000:2018 provides guidelines for identifying, analyzing, evaluating, treating, monitoring and communicating risk. It also emphasizes monitoring, review and continual improvement of risk management.
This provides an important foundation.
Predictive quality management can therefore be viewed as an evolution of risk management rather than a replacement for it.
The progression could look like this:
Risk identification
↓
Risk assessment
↓
Risk treatment
↓
Risk monitoring
↓
Trend analysis
↓
Early-warning indicators
↓
Predictive analytics
↓
Proactive intervention
The maturity of the organization increases as risk information becomes increasingly dynamic and decision-oriented.
The new role of the Quality Manager
This evolution could change the role of quality professionals.
The traditional Quality Manager may spend considerable time asking:
“Are we compliant?”
The modern Quality Manager increasingly needs to ask:
“What does the data tell us about where quality is heading?”
This requires a broader skill set.
A future-oriented Quality Manager may need to understand:
quality standards;
process management;
statistics;
data visualization;
KPI design;
risk management;
root-cause analysis;
digital systems;
AI applications;
organizational behaviour.
The Quality Manager therefore becomes not simply an inspector of past performance, but a decision-support professional for future quality performance.
Five levels of predictive quality maturity
Organizations could conceptualize their development in five stages.
Level 1 — Reactive Quality
The organization responds to defects after they occur.
Question:
What went wrong?
Level 2 — Preventive Quality
The organization identifies potential problems and establishes preventive controls.
Question:
What could go wrong?
Level 3 — Measured Quality
The organization systematically collects KPI and process-performance data.
Question:
What is happening?
Level 4 — Predictive Quality
The organization identifies trends and early-warning indicators.
Question:
What is likely to happen?
Level 5 — Quality Intelligence
The organization integrates data, risk, process performance, AI and human decision-making.
Question:
What should we do now to influence the future outcome?
This five-level model is not an ISO classification. It is a conceptual way of describing organizational maturity.
What should organizations do first?
Organizations do not need to begin by purchasing an expensive AI platform.
The first step is usually much simpler:
Improve the quality of the data.
An organization should identify:
What quality problems occur repeatedly?
What data are already being collected?
Which indicators appear before those problems?
Can the indicators be measured consistently?
What threshold should trigger investigation?
Who receives the warning?
What action should follow?
Was the intervention effective?
For example:
Defect → Cause → Precursor → Indicator → Threshold → Warning → Action → Result
This creates a feedback loop between risk management and quality management.
The future may be “quality before failure”
For decades, quality management has developed through several stages.
Quality Control
“Find the defect.”
↓
Quality Assurance
“Build a system to prevent the defect.”
↓
Quality Management
“Manage the processes that create quality.”
↓
Risk-Based Quality
“Identify what could prevent quality objectives.”
↓
Predictive Quality
“Identify the warning signs before failure.”
↓
Quality Intelligence
“Use data, technology and human judgment to influence future quality performance.”
This does not make traditional quality methods obsolete.
Inspection, testing, audits, corrective actions and root-cause analysis remain essential.
But they become parts of a larger system.
The real challenge: turning prediction into action
Predictive technology alone will not improve quality.
An organization may have an excellent dashboard showing that a problem is becoming more likely.
But if nobody acts, the prediction has little value.
This creates a new quality-management equation:
Prediction + Decision + Action = Quality Improvement
Without action, predictive analytics becomes another reporting system.
The objective should therefore not be to produce more dashboards.
It should be to produce better decisions earlier.
From “What went wrong?” to “What is the signal?”
The evolution of quality management is not about abandoning the principles that organizations have used for decades.
It is about extending them.
Risk-based thinking taught organizations to consider uncertainty before problems occur.
ISO 9001:2026 reinforces the proactive treatment of risks and opportunities while maintaining the process approach and continual improvement.
The next question is therefore increasingly data-driven:
Can we recognize the signal before the failure?
For construction companies, manufacturers, infrastructure owners and service organizations, this could become one of the defining capabilities of modern quality management.
The future Quality Manager may spend less time simply documenting what went wrong.
Instead, the quality function may increasingly become responsible for identifying where performance is heading — and helping the organization act before a quality problem becomes a quality failure.
The ultimate goal is not to predict everything.
It is to create an organization capable of seeing earlier, deciding faster and preventing better.
That is the transition from risk-based thinking to predictive quality management.
