Management

Not everything is decided in a commit: when design meets governance

Motorista diante de caminhos divergentes, com placas Design, Tecnologia, Segurança, Pessoas e Governança.

Not everyone holding the steering wheel should decide the destination alone.

In software development, a change may begin with a simple question: “Where should this button go?” It sounds like interface design. But what if that button changes access to an essential service, exposes sensitive information, or creates duties for another department? The technical question remains, but it is no longer the whole question.

When a screen decision becomes more than design

Imagine an institution building an enrollment portal. Product wants fewer fields to improve completion. Service teams need certain information to identify people requiring support. Privacy specialists question whether collecting it is proportionate to its purpose. Each has a reasonable argument; none sees the entire system alone.

Choosing a winner based on rank, confidence, or urgency is a poor answer. So is convening a committee to choose a button color. The challenge is recognizing when implementation starts defining institutional priorities, risks, and responsibilities.

A badly spaced form needs design. A form that excludes people by requiring a document they may not possess needs a broader decision. The screen is where the choice appears; its effects determine what kind of choice it is.

What if design decided to remove the brakes?

Imagine a design team deciding that a car should have no brakes because they simply do not harmonize with its appearance.

The reasoning might seem consistent from a purely aesthetic standpoint. Yet a car carries people, shares roads, and must stop safely. Removing its brakes would not be a bold design choice; it would undermine an essential function and put lives at risk.

The same applies to digital systems when beauty, simplicity, or performance eclipses security, accessibility, and rights. A solution can look perfect within one specialty and still be unacceptable in its real-world context.

Governance need not choose the brakes' color. It must ensure the vehicle can brake.

What does governance actually mean?

Governance is the system through which an organization directs decisions, assigns authority, oversees outcomes, and remains accountable. It is not synonymous with bureaucracy, unanimity, or endless meetings.

Management plans and executes work. Design translates needs and constraints into usable experiences. Governance defines who may decide what, according to which criteria, within which limits, and with which accountability. These functions overlap without being interchangeable.

ISO/IEC 38500:2024 addresses governance of organizational technology use through evaluation, direction, and oversight. It applies across public, private, and nonprofit institutions. Its value is to shift attention from the technology itself to the organization's responsible use of it.

Good governance also means clear delegation. Not every software change needs an executive decision.

What if brakes spoiled the design?

Imagine a design team deciding a car should have no brakes because they do not harmonize with its appearance.

This deliberately absurd scenario exposes a crucial distinction: removing brakes might satisfy an aesthetic preference, but it would eliminate an essential safety function. The error is treating a safety requirement as a purely visual choice.

Digital equivalents include removing a critical confirmation, hiding a necessary warning, or relaxing permissions merely to make a flow smoother. Appearance matters, but it cannot override safeguards. Governance defines these boundaries and brings relevant expertise together.

When everybody seems right

Consider an online appointment system. Experience designers want two-click bookings. Operations wants confirmation to reduce no-shows. Security asks for additional verification to prevent abuse. Accessibility specialists warn that complex verification can exclude users with disabilities.

Ease of use matters. So do service continuity, security, and accessibility. Their concerns need not cancel each other out.

Some disagreements arise not because someone is uninformed, but because each team sees a different legitimate part of the problem. Sometimes a design can improve several goals at once; sometimes trade-offs are genuine. Governance does not pretend conflict disappears. It makes criteria, impacts, and accountability explicit.

That is why absolute certainty deserves scrutiny. Confidence helps with execution; decisions affecting other people require questions too.

Five questions at the boundary

1. Who may be affected? A reversible visual preference usually belongs to design. Potential exclusion, discrimination, data exposure, or harm to well-being changes the scale of the decision.

2. Does the effect stop at the interface? A field can alter workflows, storage, authorizations, costs, and responsibilities across departments.

3. Are consequences reversible? A margin can be fixed in the next deployment. Confidential information cannot be erased from someone's memory after disclosure.

4. Are legitimate goals in conflict? Conversion versus data quality, convenience versus fraud prevention, transparency versus confidentiality. Opposing local metrics may demand a broader institutional objective.

5. Who is authorized to accept the risk? Implementers may not be legally accountable for the result, nor know the users most affected. Without clear authority, inertia makes the decision.

No single question automatically demands a committee. Together they reveal when other expertise and a designated decision-maker are needed.

Three fictional examples from digital work

1. The public enrollment list

An institution wants to publish course registrations. Web developers can easily display names and filters. Program staff value transparency. Privacy experts ask whether every detail really needs to be public.

The question becomes: what information is necessary, to whom, under what basis, and for how long? Once the institution answers, design can communicate that decision clearly.

2. A dashboard that rewards speed

A dashboard ranks departments by cases closed. Its metric is objective but may reward superficial handling of complex cases. One team wants visible targets; another argues for quality and case complexity.

The graph may be technically perfect and institutionally misguided. Before choosing bars, colors, and rankings, decide what behavior the measure should encourage and what unintended consequences should be monitored.

3. A vendor requests administrator access

A supplier requests administrator credentials to fix an urgent defect. Developers see a faster fix; security sees a least-privilege violation; operations sees costly downtime.

This is not simply “solve it” versus “block it.” Temporary access, scoped permissions, audit trails, and documented exceptional approval can reconcile the objectives. Governance sets the boundaries; engineering implements them.

Governance is neither a vote nor mandatory consensus

Listening to every relevant team does not mean granting everyone veto power. Full agreement may be impossible; urgent decisions cannot remain suspended indefinitely.

Mature organizations distinguish who recommends, who is consulted, who implements, and who decides. The decision authority considers the arguments, states the criteria, and owns residual risks. Consultation can be broad without making responsibility diffuse.

Where appropriate, people affected by the service should also be heard. Internal expertise alone may miss real-world user challenges.

A lightweight decision method

For a consequential change, write down:

  1. Problem and purpose: what are we solving, for whom, and what counts as success?
  2. Alternatives: what options exist, including doing nothing for now?
  3. Impacts and evidence: who benefits, who may lose, what is known, and what is still a hypothesis?
  4. Expertise and authority: who must be consulted, and who has the formal authority to decide?
  5. Decision and safeguards: which option, why, which controls, and who monitors the results?
  6. Review: when should the choice be revisited or reversed?

This might be a single page, an expanded architecture decision record, or a formal deliberation, depending on the stakes. Proportionality matters: the decision process should not cost more than the risk it manages.

The NIST AI Risk Management Framework uses contextual understanding, risk assessment, management, and clear governance responsibilities for AI systems. Its broader lesson here is methodological; it is not a requirement to apply an AI framework to every website.

The wheel, the map, and the passengers

The driver needs autonomy. Seeking permission for every turn would be absurd. But setting the destination, assessing road hazards, and considering passengers' needs are not always individual decisions.

Technical autonomy makes institutional software faster, better, and more innovative. Problems arise when a consequential decision is dismissed as an implementation detail—or when “governance” becomes an excuse to strip qualified teams of autonomy.

Not everything needs a committee. But not everything should be decided in a commit.

The crucial question is: are we choosing how to execute a legitimately established decision, or are we quietly deciding which interests, rights, and risks the institution will accept?

In the second case, the next step might not be opening the code editor. It might be opening a conversation.

That does not diminish the competence of whoever proposed the solution. It recognizes that organizations work better when different kinds of knowledge meet before consequences reach people.


References

EN

Ernani Nascimento

Developer, entrepreneur and business administrator. I build software and share what I learn along the way.

Meet the author ↗

Keep exploring.

Management

How to Become a Badass Leader ↗

What can Mandela, Martin Luther King Jr., Wangari Maathai, Zilda Arns, and Shackleton teach us about leadership? Practical lessons in respect, listening, developing people, and taking responsibility.

FROM IDEA TO PRACTICE

Have something in mind?
Let’s build it.

Let’s talk ↗