Advanced Project Quality Management: Perform Quality Assurance and Control Quality

Quality assurance and quality control serve different but connected purposes in a project. Assurance asks whether the methods, responsibilities and operating conditions give reasonable confidence that requirements will be met. Control examines actual work products and performance against agreed criteria. A project can have diligent inspections and still repeat the same defects because its development process is weak; it can also have excellent procedures and still release a defective product when inspection or validation misses a critical scenario.

Older PMP preparation materials sometimes use the label “Perform Quality Assurance,” while later process-oriented teaching describes “Manage Quality.” Those labels belong to the history of project-management reference frameworks, not to two separate exams available in October 2026. PMI’s July 2026 PMP assessment covers professional decisions across People, Process and Business Environment. Candidates should understand the work rather than infer that one memorized process table is the current examination blueprint.

Separate the process question from the result question

Imagine a public-sector team replacing the permit application system used by residents and building inspectors. The project promises that applications will be submitted, checked and tracked securely without losing required attachments. A quality-assurance review might ask whether the development team consistently traces requirements into test cases, whether access-control changes receive review and whether exceptions are escalated. A control activity might run actual attachments through the system and verify that permissions, calculations and confirmation notices work as specified.

Both activities are necessary. A signed procedure is not evidence that a particular attachment arrived intact. A passing functional test is not evidence that all future changes will receive proper review. The assurance system should improve how people produce dependable results; control should provide direct evidence about the results being produced.

The project quality plan must explain what the project considers good enough: applicable rules, testable acceptance criteria, who records evidence and who is authorized to accept a deviation. Without a consistent standard, audits and tests can generate activity without establishing conformance.

Build assurance into normal project work

Assurance is most useful when it shapes the workflow early. For the permit system, the project might require peer review of security-sensitive code, traceability between permit rules and implementation, accessibility checks during design, and a controlled procedure for introducing new document types. These practices reduce the chance of defects, but the project still needs evidence that they were followed effectively.

Process audits provide one type of evidence. An auditor may sample recent changes and ask whether review decisions, testing records and approval dates match the documented process. Findings might show that an informal emergency pathway bypasses the intended controls or that different teams interpret the same security rule inconsistently. The follow-up should test whether the process actually improved rather than close the audit because a policy document was reissued.

Not every assurance activity requires an external auditor or a heavyweight standard. An adaptive product team can incorporate requirement refinement, automated pipeline checks and release retrospectives. A highly regulated construction or healthcare initiative may require more formal independent verification and traceability. Tailoring matters: the controls should be proportionate to the consequences of failure.

Quality assurance should also observe feedback from quality control. If the same attachment-loss defect keeps appearing, the team should investigate the requirements and development method rather than simply add another late test. An assurance process is not a substitute for technical problem solving; it helps make effective problem solving repeatable.

Use control activities to obtain objective evidence

Control is closer to the specific deliverable. A tester may inspect a generated permit, compare its fields with approved requirements, run a realistic transaction through an integration or measure response time under a defined workload. An inspector may check whether a manufactured item conforms to tolerances. The method changes with the project, but the principle is the same: examine the output against a criterion that was agreed in advance.

Inspection, functional testing, statistical sampling and monitoring are distinct techniques. Sampling can be appropriate when examining every item would be impractical, but the selection method and residual uncertainty must be disclosed. A sample of ten uncomplicated applications cannot establish that the system handles uncommon high-risk permit types correctly. Critical failure modes may require targeted tests beyond random sampling.

When a test fails, the finding needs sufficient detail to reproduce and assess it: environment, requirement, test conditions, expected result, observed result, affected records and the potential consequence. Calling a result “failed” without preserving this context weakens both correction and learning.

The basic quality-control tools help make some patterns visible. A check sheet can collect failure counts, a Pareto chart can show concentration, a fishbone diagram can organize hypotheses, and a control chart can distinguish ordinary from unusual process behavior when its statistical assumptions hold. Choosing the wrong tool or using unreliable data may create the appearance of certainty without useful insight.

Distinguish verified, validated and accepted deliverables

Verification concerns whether a result satisfies specified requirements. Validation concerns whether it meets its intended use or need. In project practice, these questions overlap with formal acceptance but are not always answered by the same people. A technical team might verify that the permits system stores every field exactly as specified, while building inspectors discover that the workflow prevents a mandatory real-world review step. The software can satisfy a mistaken specification and still be unfit for its purpose.

Formal acceptance depends on the project’s approved criteria and decision authority. A test engineer may certify that an interface passed; that does not necessarily authorize a sponsor to accept unresolved business impacts. Conversely, a business user’s informal compliment cannot replace a required regulatory approval. Good management records what was verified, what was validated and who accepted the deliverable.

The distinction becomes especially important when changes are proposed. A defect correction might fix one calculation while affecting another approved interface. Retesting must address the relevant risks, and a change that alters scope or budget should use the project’s documented change-control authority. A ticket marked resolved is not always an accepted change.

Respond to defects and improve the system that created them

Consider an application that occasionally assigns a permit to the wrong reviewer. First, contain the active problem if possible and determine which decisions or records may be affected. Then establish whether the fault results from an incorrect routing table, missing exception handling or a confusing business rule. A plausible explanation is only a hypothesis until evidence supports it.

The immediate corrective action may be to repair routing logic and reprocess affected cases. The broader preventive action may require clarifying ownership of reviewer assignments, adding a contract test or ensuring that rule changes are reviewed across departments. Both should be assessed for effectiveness. A new procedure that nobody uses is not a successful improvement.

A distinctive quality-improvement diagnosis follows a finding through containment, investigation, controlled correction and post-change verification. That work should use the organization’s actual measures, rather than assume any improvement in one dashboard proves the entire system is better.

When a problem appears repeatedly, examine incentives as well as engineering mechanics. If managers reward only speed of release, teams may suppress test failures or avoid recording uncertainty. An assurance review should question whether such conditions undermine the stated quality policy; otherwise the same defect can recur under a different label.

Make audit evidence and quality measurements trustworthy

Useful evidence has context. An audit result should explain what was checked, which criteria applied, the sampling scope, what was observed and why the conclusion follows. A dashboard should identify the collection method and denominator. A defect rate computed from only escalated complaints is not interchangeable with the rate experienced by all users.

Likewise, a control chart’s statistically derived limits are not customer specification limits. A stable process can consistently fail a required response-time objective, and an unusual point can require investigation without proving a particular cause. Do not use a chart as an automatic decision engine when the underlying process or sampling assumptions are unsuitable.

Independent review is particularly important when findings could interrupt a politically important release. The person accountable for delivery should be able to respond to evidence without rewriting its meaning. Establish escalation and stop-work authority beforehand, with the sponsor aware of what quality exceptions demand a decision.

Adapt quality responsibilities to delivery approach

Predictive projects often schedule explicit quality reviews, test phases and acceptance gates. Iterative teams can build assurance and control into continuous integration, demonstration and operational monitoring. Hybrid programs may require both: a formal privacy approval for the program, and rapid automated checks for individual software increments. The names of the meetings differ; the need for reliable evidence does not.

People should know which activities they own. The project manager coordinates planning, visibility, decisions and risk. Technical specialists determine suitable tests. Product representatives confirm operational usefulness. Quality staff may audit the process. Authorized sponsors or customers accept deliverables under agreed conditions. No single role automatically substitutes for all the others.

For PMP study, a useful scenario question is not simply “Which process creates this output?” It is “What does the team know, what remains uncertain, what evidence is required next, and who can decide?” That logic transfers across industries and delivery methods far better than memorizing a particular era’s table of inputs and outputs.

Close the loop after release

Handover should include known residual issues, support ownership, operating metrics and a route for unexpected defects. A permit system may pass project acceptance yet need operational monitoring when a regulatory rule changes or a rare workload appears. The team should identify which control belongs to the continuing service and which assurance practices remain relevant for future projects.

Assurance and control in a patient-record migration

A clinic group is consolidating appointment records from three legacy applications. Quality assurance starts before import: define mapping rules, exception handling, sampling and peer review. An assurance review asks whether the team has a trustworthy method for preventing errors, not merely whether one export file appears correct.

Control activities test actual outputs. Reconcile counts, inspect identity matching, verify mandatory fields and examine what happens when duplicate records are encountered. A deliverable-oriented scope breakdown should include testing evidence and handover, not only installation. A low error percentage does not make an incorrect patient mapping acceptable if its consequences are serious.

Classify detected defects separately from future possibilities. A failed sample requires containment and correction, while an unobserved supplier outage may require a preventive response. The risk-and-uncertainty distinction helps managers avoid treating observed defects as speculative concerns. State what was checked, what remains unknown and who may approve any residual exception.

Good quality work connects requirements to verified outputs and acceptance. Even a passing automated suite cannot replace evidence that operations can use the migrated service and recover safely. Assurance reduces the likelihood of failure; control detects faults that remain; responsible project governance acts on that evidence before users bear the consequences.

Quality assurance helps make reliable work probable; quality control supplies evidence about results; formal acceptance establishes that authorized parties accept the outcome. Their interaction matters more than a rigid slogan that one only prevents and the other only detects. The project succeeds when requirements, processes, measurements and decisions remain connected from the first design conversation through real use.