A library can have excellent librarians, a strong collection, beautiful architecture and sophisticated technology—and still fail if nobody can answer a simple question:
Who is responsible for deciding what this institution is trying to do, what rules it follows, how its resources are allocated and what happens when difficult values collide?
That is the problem of governance.
Governance is easy to overlook because it rarely sits on a shelf or appears in a search result. Yet it determines which books are bought, which services remain free, which records are preserved, how privacy is protected, how budgets are spent, what technologies are adopted, what happens when a complaint arrives, and who can correct a decision when reality proves it wrong.
This article is part of eduKateSG’s How X Works programme and the How a Library Works series.
The shortest useful answer
Library governance works by defining the institution’s mission, assigning authority, setting policies, allocating resources, monitoring outcomes, managing conflicts and risks, preserving accountability and creating a legitimate way to change course when the environment changes.
Management asks, “How do we run the library today?” Governance asks the prior and larger questions:
- What is this library for?
- Whom does it serve?
- Which values must not be traded away casually?
- Who may make which decisions?
- Who checks those decisions?
- How are scarce resources allocated?
- How does the library respond when legitimate goals conflict?
- How can users challenge or appeal important decisions?
- What must remain true when leadership, technology or funding changes?
Governance is therefore the decision architecture above the operating system.
Mission is the first boundary condition
A library cannot govern itself coherently unless it knows what it is trying to preserve or produce.
A public library, school library, university library, national library, research institute library and special corporate library may all perform familiar library functions. Yet they serve different institutional purposes.
A public library may prioritise broad civic access, literacy, lifelong learning and community information. A university library may prioritise teaching, research and scholarly communication. A national library may carry legal-deposit and documentary-memory responsibilities. A specialist library may exist to support one profession or organisation.
Mission establishes the first filter through which later choices are judged.
Without a clear mission, every attractive initiative can look equally important. With a clear mission, governance can ask whether the initiative actually strengthens the institution’s purpose.
Mission is not a slogan if it changes decisions
A mission statement becomes governance only when it affects trade-offs.
If a library claims that equitable access is central but allocates almost all new funding to a premium service available only to a small group, mission and budget are misaligned.
If a research library claims long-term preservation responsibility but routinely cancels preservation infrastructure to protect visible short-term programmes, the mission is not controlling the resource decision.
Mission matters precisely because resources are finite. It becomes visible when the institution must choose.
Governance begins by assigning authority
Someone must have the legitimate authority to make decisions.
Depending on the institution, governance may involve a government ministry, municipal authority, school leadership, university administration, board of trustees, library board, chief librarian, director, senate committee or corporate leadership structure.
The exact form varies. The underlying problem does not.
The system must distinguish between:
- who defines institutional purpose;
- who approves major policy;
- who controls budgets;
- who hires and evaluates senior leadership;
- who manages daily operations;
- who can commit the institution contractually;
- who can override or appeal decisions;
- who is accountable when something goes wrong.
Ambiguous authority creates drift because different actors can make incompatible decisions while assuming somebody else owns the consequences.
Governance and management are different layers
Governance should not attempt to make every operational decision.
A board that decides individual shelf labels, staff rotas or routine purchases has moved too far into management. A director who changes the institution’s core mission, privacy commitments or borrowing rights without proper authority has moved too far into governance.
The distinction can be expressed simply:
- Governance: purpose, values, authority, policy, risk, accountability and strategic direction.
- Management: execution, staffing, workflow, implementation, scheduling and operational control.
The layers need each other. Governance without competent management produces intentions without capability. Management without governance can become efficient motion in the wrong direction.
Policy turns values into repeatable decisions
Libraries face recurring questions. Policy prevents every recurring question from becoming a fresh political negotiation.
Policies can govern:
- collection development;
- borrowing and circulation;
- acceptable use of facilities and networks;
- privacy and data retention;
- accessibility;
- donations;
- records management;
- special collections;
- children’s services;
- complaints and reconsideration;
- technology procurement;
- security;
- copyright and licensing;
- research data;
- AI-assisted services.
Policy is valuable because it creates consistency across time and staff.
Two similar users should not receive radically different treatment merely because they spoke to different employees on different days.
Good policy includes exceptions without becoming meaningless
Rigid policy can produce absurd outcomes when unusual cases arise.
Unlimited discretion can produce inconsistency, favouritism and poor accountability.
Strong governance defines both the rule and the legitimate exception path.
For example, ordinary loan periods may be fixed while accessibility accommodations permit extensions under clear authority. A rare-book rule may prohibit ordinary lending while allowing supervised access through a documented process.
The exception path is part of the policy, not an escape from policy.
Collection governance decides what the institution is willing to remember
Collection development appears operational when a selector chooses a title. At a deeper level, the criteria behind selection are governance decisions.
How much space should be given to popular demand? How much to long-term preservation? How should the library treat controversial works? Which languages deserve deeper coverage? How should it balance current users against future researchers? Which local publications must be retained even when use is low?
These questions involve values, not merely purchasing technique.
Governance establishes the principles. Professional selectors apply them title by title.
Intellectual freedom and collection pressure
Libraries can face pressure to add, remove, restrict or relocate material because individuals or groups strongly support or oppose it.
A governance system should define how complaints are evaluated before a controversy arrives.
That process may include written criteria, professional review, age and legal considerations, reconsideration procedures and a documented appeal route.
The purpose is not to guarantee that every disputed item remains forever. It is to ensure that decisions follow stated principles rather than immediate pressure.
Procedural integrity protects both the collection and the legitimacy of change.
Access rules are governance decisions
A library cannot make every resource equally available under every circumstance.
Rare material may require supervision. Licensed databases may be restricted to authorised users. Archives may contain confidential records. Children may have age-related service rules. High-demand material may have shorter loan periods.
Governance determines how these restrictions are justified, who can impose them and whether users have alternatives.
The key principle is proportionality: restriction should correspond to a real constraint or duty rather than becoming the default response to inconvenience.
Privacy is governed before it is secured
Library privacy is often discussed as a cybersecurity issue. Governance comes first.
The institution must decide what user data it believes it should collect, why, for how long and under what disclosure rules.
Only after those questions are answered can technology secure the resulting system.
An encrypted database containing twenty years of unnecessary reading history can be technically secure and still represent poor privacy governance.
Security protects chosen data. Governance decides whether the data should exist.
Accessibility is also a governance commitment
Accessibility cannot depend entirely on individual goodwill.
If accessible services exist only when one knowledgeable employee happens to be present, accessibility is fragile.
Governance can embed accessibility in procurement rules, building standards, digital requirements, collection choices, staff training and service policy.
That turns accessibility from an exception response into institutional capability.
Budgets reveal governance priorities
Mission becomes visible in money.
Libraries allocate funds across staff, collections, buildings, technology, programmes, preservation, accessibility, security and administration.
Every budget creates opportunity costs.
An expensive database may strengthen research but reduce funds for books. Extended opening hours may improve access but require more staffing. A major renovation may increase future capability while constraining current acquisitions.
Governance does not eliminate trade-offs. It makes them explicit, authorised and reviewable.
Capital and operating budgets solve different problems
A new building, automated sorter or repository platform may be funded as capital investment. Staff salaries, licences, subscriptions, maintenance and utilities continue every year.
Governance must consider lifecycle cost rather than purchase price alone.
A technology that is cheap to acquire and expensive to maintain can create future budget traps. A donated collection can be free to receive and costly to catalogue, house and preserve.
Every new capability is also a future maintenance obligation.
Staffing is a governance issue because capability lives in people
A library can purchase technology but cannot purchase institutional judgement in one transaction.
Cataloguing, preservation, reference, licensing, data management, accessibility and community service depend on expertise accumulated over time.
Governance therefore decides which capabilities must exist internally, which can be shared through consortia, which can be purchased from vendors and which should never be outsourced completely because they are too central to institutional responsibility.
Staffing strategy is capability architecture.
Professional judgement needs boundaries and protection
Libraries employ professionals precisely because many decisions cannot be reduced to simple rules.
A reference librarian must interpret uncertain questions. A conservator must choose among treatments. A selector must judge quality and collection fit. A metadata specialist must resolve ambiguous identities.
Governance should give these professionals enough discretion to use expertise while providing standards, documentation and review for high-impact decisions.
Micromanagement destroys expertise. Unreviewable discretion destroys accountability. Good governance designs the boundary between them.
Technology governance begins before procurement
Libraries now depend on catalogues, discovery systems, authentication, ebook platforms, databases, cloud services, analytics, self-check systems and digital repositories.
Technology decisions can quietly rewrite library policy.
A vendor platform may collect more user data than the library historically retained. A proprietary system may make records difficult to export. An ebook licence may remove traditional lending flexibility. A cloud platform may place critical services under another organisation’s outage and security model.
Technology governance therefore asks about mission fit, privacy, accessibility, interoperability, exit, data ownership, security, preservation and total cost before a contract is signed.
Vendor dependence is a governance risk
Outsourcing can create efficiency. It can also create dependency.
If one vendor controls the catalogue, authentication, analytics and digital content, the library may find it difficult to change suppliers without major disruption.
Governance therefore needs exit questions:
- Can we export our data?
- In what format?
- What happens to user data after termination?
- Do perpetual-access rights survive?
- Can another system ingest our records?
- How long would migration take?
- Which capabilities would stop immediately if the vendor disappeared?
The return path should be understood before the outward journey begins.
Risk governance: what can seriously damage the library?
Libraries face physical, financial, technological, legal and reputational risks.
- fire or flood can damage collections;
- cyberattacks can interrupt services or expose data;
- budget cuts can weaken capability;
- vendor failure can remove digital access;
- poor licensing can create legal exposure;
- staff departures can remove institutional knowledge;
- policy failures can undermine public trust;
- climate change can alter building and preservation risks.
Governance decides which risks deserve active mitigation, which can be insured, which can be shared and which must simply be accepted.
Risk management is therefore resource allocation under uncertainty.
Preservation governance asks what must survive
Preservation is costly because future users cannot pay today’s bills.
Governance has to represent future interests in present decisions.
Which materials are irreplaceable? What level of redundancy is justified? Which digital formats should be migrated? Which local publications deserve permanent retention? How much current access can be sacrificed to protect fragile originals?
These are stewardship questions that extend beyond ordinary annual performance metrics.
Buildings are governed assets
Library buildings tie the institution to long-lived capital decisions.
Renovation, expansion, storage strategy, energy systems, accessibility, flood resilience, flexible space and collection density can shape capability for decades.
Governance therefore asks not only whether a building solves today’s need but whether it preserves options for tomorrow.
Flexibility has value because forecasts fail.
Partnerships create shared capability and shared obligations
Libraries work with schools, universities, museums, archives, government agencies, publishers, technology firms, donors and other libraries.
Partnership can expand reach without requiring the library to build everything itself.
It also creates governance questions:
- Who owns jointly created data?
- Whose privacy rules apply?
- Who pays recurring costs?
- Who controls branding?
- Who preserves outputs after the project ends?
- Who carries liability?
- How can either side exit?
Partnership works when responsibilities are as clear as benefits.
Donors influence collections—and governance must preserve institutional independence
Donations can transform a library.
A donor may fund a building, collection, programme or digitisation project. But restricted gifts can also commit the institution to future expenses or conditions.
Governance should evaluate whether donor terms are compatible with mission, legal obligations, intellectual independence and long-term capacity.
A generous gift that requires decades of unsustainable maintenance may weaken the institution it intended to help.
Complaints are governance sensors
A complaint is not automatically proof that a policy is wrong.
It is evidence that the system should inspect a particular interaction.
Good complaint processes record what happened, identify the relevant rule, provide a response, permit escalation where appropriate and detect repeated patterns.
Ten unrelated complaints may be noise. Fifty similar complaints may reveal a broken workflow, inaccessible interface or policy that no longer matches reality.
Governance treats complaints as feedback without allowing every complaint to rewrite policy immediately.
Appeals make power reviewable
Whenever the library makes a consequential decision—restricting access, rejecting a donation, removing material, blocking an account, denying a request—there may be a need for review.
An appeal process does not guarantee reversal.
It ensures that power is not final merely because the first decision-maker exercised it.
Reviewability strengthens legitimacy because the institution can correct mistakes without pretending mistakes never happen.
Transparency has limits but remains essential
Users do not need access to every internal document to understand how a library is governed.
But major policies, service rules, privacy commitments, collection principles and complaint routes should be understandable.
Transparency makes rules predictable before conflict arises.
Some information must remain restricted—security details, confidential personnel matters, sensitive user data, privileged negotiations. Governance therefore distinguishes legitimate confidentiality from unnecessary opacity.
Metrics tell governance what is happening, not what should matter
Libraries can measure loans, visits, downloads, searches, programme attendance, reference transactions, digital usage, cost per use and user satisfaction.
These measurements are valuable. They are not the mission.
A rarely used local-history collection can be important. A heavily used entertainment service can be valuable without representing the whole purpose of the institution. A preservation programme may show little visible short-term output while preventing irreversible long-term loss.
Governance uses metrics as instruments, not as substitutes for judgement.
Leading and lagging indicators
Some measurements report what already happened. Others can warn what may happen next.
Circulation decline is a lagging indicator of reduced borrowing. Rising failed searches may be an earlier indicator of collection gaps. Increasing staff vacancies may warn of future service degradation. Deferred maintenance may predict building or preservation failures before an outage occurs.
Mature governance watches both outcomes and precursors.
Governance should preserve institutional memory
Leadership changes. Staff retire. Projects end.
If important decisions exist only in people’s memories, the institution can repeatedly rediscover the same problems or reverse good choices accidentally.
Decision records, policies, minutes, contracts, risk registers, architecture documents and project retrospectives preserve why choices were made.
This does not mean every old decision must remain forever. It means future leaders should be able to distinguish an intentional design from an accident.
Policy has a lifecycle
A policy written once and forgotten can become a source of institutional drift.
Technology changes. Law changes. User behaviour changes. New risks appear. Old assumptions become wrong.
Governance should therefore give policies review dates, ownership and triggers for earlier reconsideration.
The rule remains stable enough for consistency but revisable enough to remain connected to reality.
Crisis governance
Normal governance assumes time for consultation and routine approval.
Fire, cyberattack, public-health emergency, severe weather, major outage or security incident may require rapid decisions.
Crisis governance defines emergency authority in advance.
Who can close the building? Who can isolate systems? Who communicates with users? Who can authorise emergency spending? When must temporary decisions return to ordinary oversight?
Emergency power should be fast enough to protect the institution and bounded enough not to become permanent by inertia.
Continuity planning
A library is a continuity institution. Its governance should therefore plan for service continuity.
If the building closes, can digital services continue? If the catalogue fails, can staff still identify critical holdings? If one supplier disappears, is there a fallback? If key staff are unavailable, are essential procedures documented?
Continuity planning maps critical functions, dependencies, alternative routes and recovery priorities.
The question is not whether failure can be prevented completely. It is whether failure has to become institutional paralysis.
Governance across library networks
When libraries join consortia, shared catalogues or preservation networks, authority crosses institutional boundaries.
Consortium governance must define voting, cost sharing, shared standards, data rights, service expectations and exit.
The local library remains accountable to its own users while also accepting network rules.
This creates layered governance: local authority inside shared infrastructure.
Interoperability reduces governance lock-in
Standards are often treated as technical details.
They can also protect institutional freedom.
If records, identifiers and systems are interoperable, the library can change vendors, join networks and share resources more easily.
If everything depends on one proprietary format, technical design becomes a governance constraint.
Interoperability preserves options.
Public value is broader than transaction volume
A library creates value in ways that are difficult to compress into one metric.
It supports literacy, research, cultural memory, social connection, quiet study, internet access, job seeking, childhood discovery, civic information and preservation.
Governance must therefore resist the temptation to optimise the institution around whichever number is easiest to count.
A library is a portfolio of public capabilities.
User participation improves legitimacy
Users do not need to vote on every operational decision for their experience to matter in governance.
Surveys, advisory groups, consultations, usability testing, complaint data, focus groups and community partnerships can reveal where institutional assumptions differ from lived experience.
Participation works best when the library is clear about what is genuinely open to influence and what is constrained by law, budget or preservation duty.
Consultation without any possibility of effect becomes theatre. Decision-making without any listening becomes brittle.
Ethics enters where law stops
Not every difficult library decision has a complete legal answer.
A vendor practice may be lawful and still conflict with library values. A data collection may be permitted and still be unnecessarily intrusive. A controversial work may be legal and still require careful age, context or location decisions.
Governance therefore requires ethical reasoning as well as compliance.
Law defines important boundaries. Mission and professional values help decide within them.
Governance of AI in the library
AI introduces a new decision layer because systems can now classify, recommend, summarise, translate, answer questions and prioritise material automatically.
The governance question is not simply “Does the AI work?”
It is:
- What task is the AI authorised to perform?
- Which data may it see?
- May user interactions be retained or used for model improvement?
- What sources ground generated answers?
- How are errors detected and corrected?
- When must a human review the result?
- Can users tell when content is generated?
- How are bias and accessibility tested?
- Can the institution exit the AI provider without losing its records or workflows?
AI governance is library governance applied to a new mechanism.
The catalogue remains evidence underneath AI
An AI interface can make catalogues easier to navigate, but it should not dissolve the distinction between generated explanation and institutional record.
Stable metadata, authority records, identifiers, provenance and holdings remain the evidence layer beneath conversational discovery.
Governance protects that distinction because fluent output can otherwise appear more authoritative than the evidence warrants.
Automated decisions require contestability
If an automated system blocks an account, suppresses a search result, flags content or assigns an access restriction, the user may need a way to challenge the outcome.
Contestability is the ability to ask why, request review and correct error.
That principle becomes more important as systems become less transparent to ordinary users.
Automation should not make institutional power unreviewable merely because the decision arrived through software.
Governance must distinguish reversible and irreversible decisions
Some mistakes are easy to fix. Others destroy options.
Buying the wrong paperback may waste money. Discarding the only surviving local publication may destroy cultural evidence. Changing a website colour is reversible. Signing a ten-year proprietary technology contract may create deep lock-in.
Governance should demand more scrutiny when decisions are expensive, difficult to reverse or capable of creating permanent loss.
Decision process should scale with consequence.
The precaution problem
Too little caution exposes the library to avoidable harm.
Too much caution can freeze innovation.
Governance therefore needs controlled experimentation.
A new service can be piloted with a limited audience, measured, reviewed and expanded only if evidence supports it. A new AI system can begin with low-risk tasks before receiving authority over sensitive workflows.
Pilots create a reversible learning state between idea and institution-wide commitment.
Sunset clauses prevent temporary systems from becoming permanent by accident
Emergency measures, pilot programmes and temporary licences often survive long after their original reason disappears.
A sunset clause creates an explicit end or review point.
This changes the default. Instead of requiring someone to actively terminate a weak temporary programme, continuation requires evidence and renewed authority.
Sunsets are useful when uncertainty is high and institutional inertia is strong.
Governance is a feedback system
At its deepest mechanical level, library governance is feedback control.
- Mission defines a desired state.
- Policies translate that state into rules and boundaries.
- Management operates the library.
- Users, staff, finances, systems and collections produce observable outcomes.
- Metrics, complaints, audits, incidents and professional judgement reveal gaps between intended and actual performance.
- Governance decides whether the problem is execution, policy, resources, mission interpretation or external change.
- Corrective action changes budgets, policy, staffing, technology or strategy.
- The library operates again under the revised state.
A governing system that cannot observe reality will drift. A system that observes reality but cannot change policy will become brittle. A system that changes constantly without stable principles will become incoherent.
Good governance balances continuity with correction.
Accountability closes the loop
Authority without accountability is incomplete.
Accountability means decision-makers can be asked to explain what they did, under what authority, using what evidence, with what result and what correction follows if the result was poor.
This can occur through boards, audits, public reporting, institutional review, professional standards, financial controls, user appeals and legal oversight.
The purpose is not to punish every failed experiment. Institutions need room to learn.
The purpose is to prevent consequential power from becoming invisible.
The complete mechanism
- The institution defines its mission and the community or scholarly purpose it serves.
- Authority is distributed among governing bodies, leadership and professional staff.
- Policies translate mission and values into repeatable decision boundaries.
- Budgets convert priorities into actual resource commitments.
- Staffing preserves the expertise required to execute the mission.
- Collection, access, privacy, accessibility, preservation and technology decisions are made within those boundaries.
- Partnerships and vendors extend capability under explicit obligations and exit rules.
- Risk management identifies threats to people, collections, services, finances and trust.
- Metrics, audits, complaints and incidents observe the real system.
- Appeals and review make consequential decisions contestable.
- Policies and strategies are revised when evidence shows the current model no longer fits reality.
- Accountability records who decided, why, what happened and what must change next.
That is how library governance works.
A library is not sustained by books alone. It is sustained by a legitimate system for deciding what knowledge should remain reachable, what freedoms should be protected, which risks should be carried, how scarce resources should be used and who must answer when those decisions fail.
Governance is the quiet architecture that keeps the library recognisably itself while everything around it changes.