15. Sustaining Change and the Future of Transformation

Introduction: From Implementation to Institutionalization

Change is not complete when a new system is launched, a process is redesigned, or employees attend training. It becomes successful only when the intended behaviours, decisions, and performance improvements continue after the formal change programme has ended. This distinction is particularly important in Total Quality Management (TQM), where quality depends not on isolated initiatives but on the consistent integration of customer focus, process discipline, learning, prevention, and continual improvement into everyday work.

Many transformation programmes produce an initial improvement followed by regression. Employees return to familiar routines, managers stop reviewing the new measures, incentives reward short-term output rather than quality, and newly acquired knowledge is not incorporated into procedures or training. Such regression is not necessarily caused by employee resistance. It may indicate that the organization changed its project but not its operating system.

Sustaining change therefore requires institutionalization. Institutionalization occurs when a new way of working is embedded in structures, systems, leadership routines, performance measures, resource decisions, social norms, and organizational learning. It also requires benefits realization: the deliberate process of verifying whether the expected operational, financial, customer, employee, and strategic benefits have actually been achieved.

This chapter examines how organizations sustain change, conduct post-change reviews, develop resilience and adaptive capacity, and respond to emerging practices such as digital transformation, data-enabled quality management, artificial intelligence, and continuous transformation. The central argument is that sustainable transformation is less a final stage than an organizational capability.

Learning Objectives

After studying this chapter, students should be able to:

1. Explain why change initiatives regress and distinguish implementation from institutionalization.

2. Analyse how systems, leadership, incentives, culture, and learning sustain new behaviours.

3. Design a benefits-realization and post-change review process.

4. Evaluate organizational resilience and adaptive capacity.

5. Apply frameworks for sustaining quality-related transformation.

6. Assess the opportunities and risks associated with digital, data-driven, and AI-enabled change.

7. Recommend managerial and governance practices for responsible and durable transformation.

The Nature of Sustainable Change

Sustainable change has three related dimensions. The first is behavioural sustainability: people consistently perform the new activities and make decisions according to the intended principles. The second is performance sustainability: the improvement remains visible in relevant outcomes rather than disappearing after initial attention declines. The third is institutional sustainability: the organization’s structures and routines support the new state even when particular sponsors, consultants, or project leaders leave.

A useful distinction can be made between adoption, adherence, performance, and renewal. Adoption means that individuals begin using the new practice. Adherence means that they use it reliably. Performance means that the practice produces the intended results. Renewal means that the organization continues to refine the practice as circumstances change. A process may be adopted without being followed, followed without producing value, or successful for a period without being renewed. Sustainable transformation must address all four conditions.

In TQM, sustainability is closely connected to process ownership. If a redesigned process belongs only to a temporary project team, accountability becomes ambiguous after implementation. The process owner must have authority, performance information, improvement resources, and responsibility for maintaining process capability. Quality is sustained when control and improvement are connected: control prevents unacceptable variation in the current process, while improvement raises the process to a better level of performance.

Why Improvements Regress

Regression commonly arises from several interacting causes:

• senior leaders withdraw attention once the launch is completed;
• performance measures continue to reward old behaviours;
• workload pressures cause employees to bypass the new process;
• training is delivered once but not reinforced through coaching;
• procedures are documented but inconsistent with information systems;
• improvement targets are imposed without adequate resources;
• local adaptations are treated as non-compliance even when they reveal process weaknesses;
• benefits are claimed without establishing a baseline or confirming causation;
• lessons remain with the project team instead of entering organizational knowledge.

These causes demonstrate that sustainability is a system problem. Asking employees simply to “maintain the change” is inadequate if the surrounding system continually encourages the previous state.

The Architecture of Institutionalization

Institutionalization can be understood as the alignment of six organizational mechanisms: strategic intent, work systems, people systems, measurement, leadership routines, and learning. They should reinforce one another rather than operate as separate interventions.

Strategic Intent and Governance

A change becomes durable when it is connected to strategic priorities and governance processes. The organization should define why the change matters, which outcomes it supports, and who is accountable for its continuation. Governance should not mean excessive bureaucracy. It should provide decision rights, escalation routes, resource oversight, risk review, and benefit accountability.

A transformation charter may identify the executive sponsor, process owner, benefit owners, affected stakeholders, target outcomes, assumptions, and review dates. After implementation, governance should evolve from project control to operational stewardship. The steering committee may be dissolved, but accountability must be transferred explicitly to line management, process councils, quality boards, or other permanent structures.

Work Systems and Standardization

Standardization is not the mechanical imposition of one method in every circumstance. It establishes the current agreed method, critical controls, responsibilities, and conditions under which deviation is permitted. Standard work is valuable because it makes variation visible and provides a stable basis for improvement.

Effective standardization links four elements:

1. documented process requirements;

2. employee competence to perform the process;

3. system and equipment configuration that supports the process; and

4. monitoring that reveals whether the process is functioning as intended.

Procedures should be accessible at the point of work and revised through controlled change processes. Obsolete forms, duplicated databases, and conflicting instructions are frequent sources of regression. In regulated or safety-critical industries, document control and traceability also have legal and professional significance.

Leadership Routines and Visible Reinforcement

Leadership sustains change through repeated attention rather than occasional speeches. Useful routines include reviewing leading indicators, asking how a process is performing for customers, visiting the workplace, recognizing desired behaviours, removing barriers, and allocating resources to corrective action.

Leaders must also demonstrate consistency. If executives endorse quality but reward only volume, employees receive a stronger message from the reward system than from formal communication. Visible reinforcement is most credible when leaders accept inconvenient information, investigate systemic causes, and avoid blaming individuals for problems generated by the process.

Incentives and Recognition

Incentives influence what employees regard as important. A sustainable reward system should balance results with the behaviours and capabilities that produce those results. For example, rewarding only the number of cases processed may encourage premature closure, whereas a balanced system may include accuracy, customer outcome, cycle time, compliance, teamwork, and improvement contribution.

Incentives must be designed carefully. Individual competition can undermine cross-functional quality improvement; financial bonuses can encourage manipulation of measures; and recognition programmes can become symbolic if employees lack time or authority to improve work. Recognition may be monetary or non-monetary, but it should be timely, credible, and connected to values the organization wishes to institutionalize.

Capability, Socialization, and Learning

Training is necessary but insufficient. Capability is sustained through induction, role-specific learning, coaching, communities of practice, peer review, simulation, and access to expert support. New employees should learn not only the formal procedure but also the quality rationale behind it. Understanding why a control exists increases the likelihood that employees will protect it under pressure.

Socialization is equally important. Informal norms influence whether people report problems, challenge unsafe practices, and share knowledge. A culture of learning treats defects and near misses as information for improvement while maintaining accountability for negligence or deliberate misconduct. Psychological safety does not mean the absence of standards; it means that people can raise concerns and admit uncertainty without disproportionate interpersonal punishment.

Measurement and Feedback

Measurement sustains change by making performance visible, but poorly designed measures can destroy it. A balanced measurement architecture should include:

• outcome indicators, such as customer satisfaction, defect rates, cost, and reliability;
• process indicators, such as first-pass yield, response time, or adherence to critical controls;
• leading indicators, such as training completion, improvement activity, risk reviews, and unresolved barriers;
• balancing measures, which reveal unintended consequences, such as employee workload or service accessibility.

Measures should have clear definitions, owners, data sources, review frequency, and decision rules. Statistical process control may be used where data and process conditions are appropriate. A single adverse result should not automatically trigger managerial intervention if it represents common-cause variation; conversely, a stable average may conceal unacceptable variation among customer groups or locations.

Frameworks for Sustaining Transformation

The Reinforcement Perspective

A practical reinforcement cycle consists of five activities: confirm the desired behaviour, observe its use, provide feedback, correct barriers, and recognize sustained performance. The cycle is continuous rather than a one-time communication event. It is especially useful during the period when formal project resources decline but old habits remain attractive.

The McKinsey 7-S Interpretation

The 7-S framework—strategy, structure, systems, shared values, style, staff, and skills—helps diagnose whether the organization is aligned with the desired state. For example, a quality transformation may fail if strategy emphasizes customer reliability but systems measure only throughput, structures isolate departments, and staff lack data-analysis skills. The framework is valuable because it prevents managers from treating change as a training or communication problem alone.

The ADKAR Continuity Lens

The ADKAR model describes individual change through awareness, desire, knowledge, ability, and reinforcement. For post-change management, reinforcement is the most direct concern, but it depends on the earlier conditions. An employee cannot sustain a practice that was never understood, or use a process for which the organization has not provided ability and resources. Managers can use the model diagnostically: a compliance problem may actually be a knowledge, system, or motivation problem.

The Learning-Loop Perspective

Single-loop learning corrects performance within existing assumptions; double-loop learning questions the assumptions, rules, or objectives that created the problem. For example, adjusting staff schedules to reduce waiting time is single-loop learning. Asking whether the service should be redesigned around demand patterns is double-loop learning. Sustainable organizations use both. They stabilize essential processes while remaining willing to revise the design when evidence shows that the underlying assumptions are no longer valid.

Benefits Realization: Demonstrating Value Rather Than Activity

Benefits realization converts a change promise into a managed set of outcomes. A benefit is a measurable improvement resulting from a change, while an output is something the project produces, such as a software module, procedure, or training programme. Confusing outputs with benefits is a common governance failure.

A benefits-realization plan should specify:

• the expected benefit and its relationship to strategic objectives;
• the baseline and target value;
• the benefit owner;
• the measures and data source;
• the date by which the benefit should appear;
• dependencies and assumptions;
• risks to realization; and
• the method for confirming attribution or contribution.

Benefits may be financial, operational, customer-related, employee-related, risk-related, or strategic. Some are tangible, such as reduced rework cost. Others are less easily monetized, such as improved trust, resilience, accessibility, or regulatory confidence. The absence of a simple monetary measure does not make a benefit unimportant; it requires an explicit and credible measurement approach.

Attribution should be treated cautiously. If defect rates improve after a new quality system is introduced, other influences may include product mix, market demand, staffing changes, or supplier performance. A benefits review should therefore examine the contribution of the change using comparative data, trend analysis, control groups where feasible, and informed stakeholder judgement. Claims should distinguish achieved benefits from forecast benefits and from benefits that remain plausible but unverified.

Post-Change Reviews and Organizational Learning

A post-change review is a structured examination of what happened, what value was achieved, what risks remain, and what should be learned. It should not be reduced to a ceremonial project closure meeting. Timing matters: an early review may assess adoption and immediate operational issues, while a later review can assess sustained outcomes and unintended effects.

A rigorous review normally considers four questions:

1. Was the intended change adopted and embedded?

2. Were the expected benefits achieved, for whom, and under what conditions?

3. What unintended consequences, residual risks, or new dependencies emerged?

4. What should be standardized, redesigned, transferred, or stopped?

Evidence may include performance data, audits, customer feedback, interviews, observation, financial analysis, incident records, and employee experience measures. The review should include multiple perspectives, especially those of users and groups who may have experienced disadvantages. Findings should be converted into named actions with owners and dates. Otherwise, the review creates knowledge without organizational learning.

A mature organization maintains a change repository containing decision rationales, assumptions, metrics, lessons, reusable designs, and known failure modes. This avoids repeating mistakes and improves the quality of future transformation planning. However, repositories should be curated; storing large volumes of unexamined documents is not the same as learning.

Resilience and the Adaptive Organization

Resilience is the capacity to prepare for disruption, absorb shocks, continue essential functions, recover effectively, and adapt on the basis of experience. It differs from simple robustness. Robustness attempts to resist disturbance through strength and control; resilience also includes flexibility, improvisation, and renewal.

For a quality-oriented organization, resilience includes operational continuity, supply-chain awareness, information integrity, workforce capability, leadership succession, and the ability to detect weak signals. It is strengthened when processes are understood rather than merely documented, when critical knowledge is distributed, and when teams practise responses to plausible disruptions.

Adaptive organizations exhibit several characteristics. They maintain a clear purpose while allowing local decision-making; use data without becoming dependent on rigid targets; combine standardization of critical controls with experimentation in improvement; develop cross-functional networks; and treat environmental change as a reason to learn rather than merely a reason to defend existing plans.

Adaptation must nevertheless be governed. Excessive local variation can undermine safety, fairness, and comparability. The managerial task is to define non-negotiable principles and controls while allowing appropriate discretion in methods. This is sometimes described as disciplined autonomy.

Digital Transformation, Analytics, and Artificial Intelligence

Digital tools can support sustainability by embedding prompts, workflow controls, real-time dashboards, automated alerts, digital standard work, and traceable approvals. Process-mining tools can reveal how work actually flows rather than how it is assumed to flow. Predictive analytics may identify patterns associated with defects, equipment failure, delays, or customer attrition.

Artificial intelligence can assist with anomaly detection, document classification, root-cause exploration, forecasting, knowledge retrieval, and personalized learning. These applications can strengthen quality improvement when they augment professional judgement and are integrated into accountable processes. They do not eliminate the need for process understanding, data governance, or human oversight.

Several risks require attention. Algorithms may reproduce historical bias, generate unreliable recommendations, expose confidential information, or encourage automation of a poorly designed process. Data drift can reduce model performance as operating conditions change. Employees and affected stakeholders should know, where appropriate, how automated decisions influence them, what review mechanisms exist, and who is accountable for errors. AI governance should address validity, explainability, privacy, cybersecurity, access control, monitoring, and the possibility of human override.

The principle “automate only after understanding and improving the process” remains important. Digitalization can make waste faster and less visible if it merely transfers an inefficient process into software.

An Illustrative Case: Sustaining a Hospital Medication-Safety Change

Background

The following fictional case illustrates the application of sustainability principles. A regional hospital introduced barcode medication administration to reduce errors involving patient identity, medicine selection, and dosage. The project included new scanners, software integration, staff training, and revised administration procedures.

Situation and Problem

During the first three months, reported administration errors declined. Six months later, scanning compliance fell in two wards. Nurses reported slow devices, incomplete patient wristbands, and difficulty scanning during emergency situations. Supervisors focused on compliance percentages, while staff began using workarounds. The project team argued that the system had been successfully implemented because the technology was operational.

Analysis

A post-implementation diagnostic distinguished adoption from institutionalization. The technology had been adopted, but the work system did not reliably support adherence. Measures emphasized scanning rates but did not include device availability, wristband quality, exception patterns, workload, or near misses. Training had occurred before launch but was not part of induction. The incentive climate implicitly prioritized speed during peak periods. No permanent process owner had been assigned after the project team was disbanded.

Approach

The hospital appointed a medication-safety process owner and established a cross-functional review group involving nurses, pharmacists, information-technology staff, infection-control specialists, and patient representatives. The group:

• analysed exception data and observed work at different times;
• repaired or replaced unreliable devices;
• introduced a rapid escalation route for missing wristbands and system failures;
• revised training to include simulation of urgent and unusual cases;
• clarified when a controlled manual procedure was permitted;
• added balancing measures for workload and delays;
• included medication-safety performance in ward review meetings; and
• conducted monthly benefits reviews rather than relying solely on the original project report.

The hospital also used a learning approach to distinguish deliberate disregard from reasonable adaptation to a defective system. Staff were expected to follow critical controls, but reporting a barrier was treated as a contribution to safety improvement.

Outcome

Over subsequent review periods, scanning compliance improved and workarounds became less frequent. More importantly, the hospital identified that the original measure had overstated sustainability. The revised system provided a fuller picture of process reliability and made ownership permanent. The organization also applied the learning to later digital projects by requiring a named operational owner, a benefits map, and a six-month post-implementation review.

Learning Points

The case demonstrates that technology is not the same as transformation. Sustainable change required alignment among equipment, procedures, capability, measures, leadership, and learning. It also shows why a low compliance result should be investigated systemically rather than interpreted automatically as employee resistance.

Challenges, Ethics, and Governance

Sustaining change creates difficult managerial choices. Standardization may improve reliability but reduce local discretion. Automation may improve consistency while creating surveillance concerns. Performance measurement may increase accountability but also encourage gaming or unfair comparison. Benefits realization may expose that an influential programme did not deliver its promised value.

Ethical governance requires honesty about results, respect for affected stakeholders, protection of personal data, fair distribution of benefits and burdens, and meaningful avenues for challenge. Leaders should not conceal negative outcomes to protect the reputation of a project sponsor. Nor should they use transformation language to justify unreasonable workloads or the elimination of professional judgement.

Common mistakes include declaring victory too early, treating resistance as a personal defect, measuring activity instead of outcomes, ignoring frontline knowledge, confusing documentation with control, and failing to retire obsolete systems. Another mistake is perpetual change without stabilization. Organizations can exhaust employees when every process is labelled urgent and no time is provided to absorb learning. Transformation requires rhythm: periods of experimentation, implementation, stabilization, review, and renewal.

Emerging Directions in Change Management

The future of transformation is likely to be characterized by continuous, data-informed adaptation rather than occasional large projects. Several developments are significant:

• transformation portfolios will increasingly be managed together to identify dependencies and cumulative employee impact;
• digital twins and process simulation may help test process changes before full deployment;
• real-time operational data will support earlier detection of drift and emerging risk;
• employee experience analytics may reveal friction in systems and workflows;
• sustainability, climate risk, and social responsibility will become more integrated into transformation decisions;
• AI-supported change planning will increase the speed of analysis but also the need for strong validation and accountability; and
• organizations will place greater emphasis on resilience, not merely efficiency.

These developments do not replace foundational management disciplines. Clear purpose, process ownership, stakeholder participation, evidence, ethical judgement, and learning remain essential. The future organization is not one that changes constantly without direction; it is one that can change deliberately without losing control of quality, values, or trust.

Best-Practice Guidance

Managers seeking to sustain a major change should ensure that:

1. the desired outcomes and baseline are defined before implementation;

2. ownership is transferred from the project to permanent operations;

3. procedures, technologies, measures, skills, and incentives are aligned;

4. leaders reinforce the change through regular operating routines;

5. performance measures include outcomes, leading indicators, and balancing measures;

6. benefits are reviewed after sufficient time has passed to test durability;

7. frontline feedback and exceptions are treated as evidence for improvement;

8. lessons are converted into accessible organizational knowledge;

9. critical controls are standardized while appropriate local adaptation is allowed;

10. digital and AI applications are governed for validity, fairness, privacy, security, and accountability.

Chapter Summary

Sustainable transformation extends beyond implementation. It requires new behaviours and results to become part of the organization’s normal operating system. Institutionalization depends on aligned strategy, governance, process ownership, standard work, leadership routines, incentives, capability development, measurement, and learning. Benefits realization provides the discipline to distinguish outputs from actual value and to verify whether benefits are achieved and sustained. Post-change reviews should examine adoption, performance, unintended consequences, and transferable learning.

Resilient organizations combine reliable controls with adaptive capacity. They prepare for disruption, learn from weak signals, distribute knowledge, and permit disciplined autonomy. Digital tools, analytics, and AI can strengthen sustainability when they are applied to understood processes and governed responsibly. The enduring lesson is that change management is not finished at project closure; it becomes an organizational capability through reinforcement, evidence, reflection, and renewal.

Key Terms

Institutionalization: The embedding of new behaviours and practices into an organization’s permanent systems, routines, and norms.

Adoption: The initial acceptance and use of a new practice, system, or behaviour.

Adherence: Consistent use of a required practice under normal operating conditions.

Benefits realization: The planned identification, measurement, ownership, and verification of benefits arising from change.

Benefit owner: The person accountable for achieving and sustaining a specified benefit.

Output: A product or deliverable produced by a change project.

Outcome: A change in performance, behaviour, capability, or condition resulting from an output.

Process owner: The person accountable for the performance and improvement of an end-to-end process.

Standard work: The current agreed method for performing a process, including essential controls and responsibilities.

Leading indicator: A measure that provides early evidence about future performance or the likelihood of achieving an outcome.

Balancing measure: A measure used to detect unintended adverse effects of an improvement.

Post-change review: A structured assessment of adoption, benefits, risks, outcomes, and lessons after implementation.

Organizational learning: The process through which experience and evidence alter organizational knowledge, routines, or assumptions.

Single-loop learning: Correcting performance without questioning the underlying assumptions or governing rules.

Double-loop learning: Revising the assumptions, objectives, or rules that shape performance.

Resilience: The capacity to prepare for, absorb, recover from, and adapt to disruption.

Adaptive capacity: The ability to adjust structures, decisions, and practices in response to changing conditions.

Disciplined autonomy: Local discretion exercised within clearly defined organizational principles and critical controls.

Process mining: The use of event data to discover, analyse, and monitor actual process flows.

Data drift: Change in the characteristics of data that can reduce the validity or performance of an analytical model.

Review Questions

Short Answer Questions

1. Distinguish adoption, adherence, performance, and renewal in sustainable change.

2. Why is process ownership important after a change project ends?

3. What is the difference between an output and a benefit?

4. Define a balancing measure and provide one example.

5. How does double-loop learning differ from single-loop learning?

Long Answer Questions

1. Explain how leadership routines, incentives, measurement, and learning combine to institutionalize change.

2. Describe a comprehensive benefits-realization process for a TQM transformation.

3. Discuss the role of post-change reviews in organizational learning and governance.

4. Explain the characteristics of a resilient and adaptive organization.

5. Critically evaluate the opportunities and risks of using AI and analytics to sustain organizational change.

Analytical / Application Questions

1. A manufacturing company’s defect rate improves for three months after a lean implementation but then returns to its previous level. Diagnose the likely institutionalization failures and propose a sustainability plan.

2. A public-service agency wants to automate decisions using historical customer data. Design a governance approach addressing validity, fairness, privacy, human oversight, and benefits realization.

3. A hospital reports high compliance with a new digital safety procedure, but frontline employees report delays and workarounds. Identify the additional measures and review methods required to determine whether the change is genuinely sustainable.

22. Multiple Choice Questions with Answers

Question 1

Which condition best indicates that a change has been institutionalized?

A. The project team has completed all scheduled activities.

B. Employees attended the required training programme.

C. The new practice is supported by permanent systems, ownership, measures, and routines.

D. Senior management announced the change publicly.

Correct Answer: C. The new practice is supported by permanent systems, ownership, measures, and routines.

Question 2

Which of the following is a balancing measure?

A. Number of employees trained

B. Reduction in defect rate

C. Percentage of processes documented

D. Employee workload after a productivity intervention

Correct Answer: D. Employee workload after a productivity intervention

Question 3

A project delivers a new software system, but customer waiting time does not improve. What does this most clearly demonstrate?

A. The output has been delivered without achieving the intended benefit.

B. The benefit has automatically been realized.

C. The project requires no further review.

D. The software system is necessarily defective.

Correct Answer: A. The output has been delivered without achieving the intended benefit.

Question 4

Which action best supports benefits realization?

A. Assigning benefit ownership only to the project manager

B. Defining a baseline, target, measure, owner, and review date

C. Measuring only financial outcomes

D. Closing the project immediately after implementation

Correct Answer: B. Defining a baseline, target, measure, owner, and review date

Question 5

In the hospital case, declining scanning compliance was interpreted most appropriately as:

A. Proof that employees opposed technology.

B. Evidence that all training should be cancelled.

C. A possible symptom of system, workload, and process-design problems.

D. A reason to remove all local discretion.

Correct Answer: C. A possible symptom of system, workload, and process-design problems.

Question 6

What is the principal difference between robustness and resilience?

A. Robustness concerns resistance, while resilience also includes recovery and adaptation.

B. Robustness requires learning, while resilience prohibits standardization.

C. Resilience applies only to technology, while robustness applies only to people.

D. There is no meaningful difference between them.

Correct Answer: A. Robustness concerns resistance, while resilience also includes recovery and adaptation.

Question 7

Which example represents double-loop learning?

A. Correcting a data-entry error using the existing procedure

B. Increasing inspection frequency without changing the process design

C. Questioning whether the current service model causes the recurring delay

D. Repeating the same training after a missed target

Correct Answer: C. Questioning whether the current service model causes the recurring delay

Question 8

Why can automation fail to sustain change?

A. Automation always reduces process visibility.

B. It may embed an inefficient process and create new risks without proper governance.

C. Automated systems cannot collect performance data.

D. Automation eliminates the need for process ownership.

Correct Answer: B. It may embed an inefficient process and create new risks without proper governance.

Question 9

Which governance principle is most appropriate for AI-supported change decisions?

A. Allow the model to operate without human review once deployed.

B. Ignore data drift if initial accuracy is high.

C. Ensure accountability, monitoring, privacy protection, and appropriate human override.

D. Use historical data without assessing bias.

Correct Answer: C. Ensure accountability, monitoring, privacy protection, and appropriate human override.

Question 10

A manager rewards only the number of cases processed, and quality falls. What sustainability lesson is most relevant?

A. Measures and incentives can reinforce undesirable behaviour if they are not balanced.

B. Quality cannot be measured in service organizations.

C. Employees should receive no performance feedback.

D. Standardization is always harmful.

Correct Answer: A. Measures and incentives can reinforce undesirable behaviour if they are not balanced.