Minimum Viable Friction: A Practical Framework for Better Decisions
Most organisations have spent years removing friction so they can move faster. The problem is that some friction was doing useful work. It created the pause where judgement, assumptions and consequences could be examined before a decision became hard to reverse.
Some friction was doing useful work
Over the last decade, organisations have worked hard to remove friction from how they operate.
Approval chains were shortened. Processes were streamlined. Decisions were pushed closer to delivery teams. Automation promised speed, consistency and scale. AI now promises to make that even faster.
Much of this work was overdue.
Excessive friction does slow organisations down. It hides accountability, creates bottlenecks and makes people wait for permission when they already have enough information to act.
But in many cases, friction was removed without asking what role it was playing.
Along the way, many organisations also removed the moments where thinking, judgement and reflection used to occur.
Key point: Minimum Viable Friction is not about slowing work down. It is about adding just enough pause at the right decision points to protect judgement, accountability and decision quality.
When efficiency undermines decision quality
Most organisational failures are not caused by poor intent or lack of capability.
They happen when decisions are made quickly on the basis of incomplete understanding.
Assumptions remain implicit. Risks feel obvious until they materialise. Second-order impacts surface only after commitments have been made. People move forward because the process allows them to, not because the decision has been properly tested.
Highly optimised systems are excellent at executing known patterns. They are far less effective at questioning whether those patterns still apply.
When everything is designed to move forward smoothly, there are fewer natural checkpoints where direction itself is examined.
Speed becomes a proxy for certainty.
This is particularly dangerous in organisations adopting AI, automation and data-driven decision support. When tools generate summaries, forecasts, recommendations or classifications quickly, the output can feel more reliable than it actually is.
This is why Changeable connects AI strategy, AI governance, data models and decision design. Faster decisions are not automatically better decisions.
Speed is useful when the pattern is known. It is dangerous when the assumptions have not been examined.
What Minimum Viable Friction means
Minimum Viable Friction refers to the smallest amount of intentional resistance needed to improve decision quality without creating unnecessary drag.
It is not about adding layers of process. It is not about returning to slow approval chains. It is not about making teams ask permission for routine work.
The intent is to introduce carefully placed pauses at the points in a decision flow where mistakes are hardest to reverse.
Those pauses do not exist to seek permission or enforce compliance. They exist to make thinking explicit.
When someone has to explain why a decision makes sense, what assumptions it relies on or how risk has been weighed, weak reasoning tends to surface quickly. Strong reasoning becomes clearer too.
This approach works because it is proportional.
Friction appears where uncertainty, impact or irreversibility are high, and stays out of the way everywhere else.
Useful distinction: Bad friction blocks work. Good friction protects judgement. Minimum Viable Friction is about knowing the difference.
Why this matters now
Several changes have made the absence of friction more risky than it appears.
Teams operate under sustained cognitive load, surrounded by dashboards, alerts, competing priorities and constant pressure to respond quickly. Reflection is often the first casualty.
At the same time, AI systems now generate recommendations, summaries, classifications and forecasts at speed. The risk is not that these outputs are always wrong. The risk is that they feel authoritative enough to bypass judgement.
Organisations are also more interconnected than ever. Decisions made in one area can ripple across systems, teams, incentives, customers and regulatory obligations in ways that are not immediately visible.
Faster decisions reduce the opportunity to notice those effects in advance.
Removing all friction in this environment does not create clarity. It creates blind spots.
The New Zealand Government’s Public Service AI Framework is a useful public-sector example of why responsible AI use requires human accountability, transparency and trust, not just faster execution.
The same principle applies outside government. Any organisation using AI or automation to support decisions needs to decide where human judgement must remain visible.
Where friction belongs
Not all work benefits from friction.
Routine, repeatable execution should be smooth and predictable. Payroll, standard reporting, infrastructure scaling, basic administration and low-risk operational processes often work best when resistance is low.
Friction belongs at inflection points.
These are moments where decisions are ambiguous, irreversible or carry material consequences for people, trust, service quality, finances or risk exposure.
The failure mode is applying friction everywhere, or nowhere.
Minimum Viable Friction is about being selective and deliberate.
What Minimum Viable Friction looks like on the ground
In practice, Minimum Viable Friction is usually light-touch.
It might take the form of a short written statement explaining what would need to be true for a decision to fail.
It might require a brief justification when accepting an AI-generated recommendation.
It might introduce a deliberate pause before committing to a non-urgent but high-impact change.
It might ask a decision owner to record the assumptions being relied on before a project moves from discovery into delivery.
These interventions are not designed to block progress. They exist to surface thinking that would otherwise remain implicit.
The value lies less in the artefact produced and more in the act of explanation itself.
This is why Minimum Viable Friction works well with AI use case discovery. Before a team invests in a tool, builds an agent or automates a workflow, it creates a short moment to test whether the decision is grounded in real value, clear assumptions and known risks.
What it is not
Minimum Viable Friction is often mistaken for distrust or an attempt to control decision-making.
It is neither.
It does not exist to protect legacy roles. It does not exist to slow down change. It does not exist to make leaders feel safer by adding another form.
It reflects a simple reality: confidence and correctness are not the same thing.
Systems that remove every pause make that difference harder to detect.
Poorly designed friction
Creates compliance theatre, delays routine work and makes people perform paperwork without improving judgement.
Well-designed friction
Creates clarity, makes assumptions visible and protects decisions that are hard to reverse.
This distinction matters when organisations introduce workflow automation or AI agents. The point is not to make every automated step slower. The point is to understand where a human decision still needs to be explicit.
A more useful definition of organisational maturity
Maturity is often equated with speed.
Faster delivery. Faster decisions. Faster change. Faster implementation.
In practice, mature organisations are those that understand when speed is appropriate and when it is not.
They know which decisions can safely move quickly, and which require more thought before commitment.
They treat judgement as a finite resource and design their systems to protect it.
Minimum Viable Friction provides a way to do this without reverting to bureaucracy. It recognises that some resistance is not a flaw in the system, but a feature that keeps decisions grounded and accountable.
This aligns with the logic behind the New Zealand Treasury’s Better Business Cases approach, which uses structured thinking to help decision-makers assess value, achievability and risk before committing to investment.
Mature organisations do not move fast all the time. They know when speed creates value and when it hides risk.
The Minimum Viable Friction framework
The framework itself is intentionally simple.
It is designed to be applied inside existing decision-making processes, not layered on top of them.
Minimum Viable Friction starts by identifying decisions that meet at least one of three criteria:
Routine operational decisions do not enter the framework.
Once a decision qualifies, three lightweight checkpoints are introduced.
Checkpoint 1: Assumption check
Before the decision proceeds, the decision owner articulates the key assumptions the decision relies on.
This does not need to be exhaustive. The goal is to surface what must be true for the decision to succeed.
The assumption check is useful because weak decisions often look strong until the underlying assumptions are named.
Checkpoint 2: Impact scan
The decision owner considers who and what will be affected beyond the immediate scope.
This includes downstream teams, customers, suppliers, systems, incentives, regulatory obligations, privacy, trust and service quality.
The intent is not to predict every outcome. It is to avoid narrow optimisation.
This is especially important for AI and automation because a change that looks efficient in one workflow can create hidden work, identity risk or quality issues elsewhere.
Useful distinction: A decision can optimise one workflow while making the wider system worse. The impact scan helps catch that before the change becomes embedded.
Checkpoint 3: Judgement statement
The decision owner briefly records why they believe this is the right decision now, given the available information, and what would cause them to revisit it later.
This can be a paragraph. It does not need to be a report.
The statement should answer:
These checkpoints can be completed in minutes. They can be written, verbal or embedded into existing tooling.
What matters is that the thinking happens and is visible.
Practical rule: If a decision is easy to reverse, low-risk and well understood, keep it moving. If it is hard to reverse, uncertain or high-impact, add just enough friction to make judgement visible.
How to use the framework
Start by mapping your current decision points.
Identify where irreversible or high-impact decisions are being made quickly, informally or implicitly.
Then introduce the framework only at those points.
Do not apply it universally. Overuse will dilute its value.
Assign a clear decision owner
Minimum Viable Friction increases accountability by making it explicit who is exercising judgement.
That does not mean one person has all the answers. It means someone is responsible for the reasoning, the trade-off and the point at which the decision moves forward.
Keep the artefacts lightweight
A paragraph is usually sufficient.
The purpose is not documentation for its own sake. The purpose is clarity in the moment.
If the artefact becomes too heavy, people will avoid it or turn it into compliance theatre.
Review outcomes periodically
If decisions consistently require rework or reversal, increase friction slightly.
If decisions move smoothly with good outcomes, friction may already be sufficient.
The framework should evolve with the organisation. As confidence, capability and shared understanding grow, the amount of friction required often decreases.
Minimum Viable Friction is not a control mechanism. It is a design tool for protecting judgement at the moments where it matters most.
How Minimum Viable Friction applies to AI decisions
AI makes Minimum Viable Friction more important because AI systems can create confident outputs very quickly.
That speed can be useful. It can also make poor reasoning harder to notice.
Minimum Viable Friction can be applied to AI decisions such as:
In these cases, the friction may be as simple as asking what the AI output is being used for, what data or knowledge source it relies on, what limitations are known, who reviews the output, who remains accountable and what would trigger escalation.
This connects directly to AI governance. Governance should not simply state principles. It should help people know where judgement must be visible before action is taken.
How Changeable applies Minimum Viable Friction
At Changeable, Minimum Viable Friction is applied as a practical decision design tool, not as a generic governance overlay.
We embed it directly into existing delivery, transformation and AI-enabled workflows so it supports momentum rather than competing with it.
In discovery and strategy work, the framework is often useful for operating model changes, technology selections, AI use-case prioritisation and investment sequencing.
In delivery and change programmes, it is useful when moving from pilot to scale, shifting ownership from project to operations or automating decisions that were previously human-led.
In AI and automation work, it helps clarify why an output is being trusted, where its limits are and what signals would trigger review.
The goal is not control. It is decision confidence that holds up under pressure.
For high-stakes decisions, Changeable’s sister practice Ministry of Insights can also help organisations simulate how a decision may land before they commit.
What Changeable helps with
Changeable helps organisations design better decision systems for AI, automation, transformation and operational improvement.
Start with a Decision Clarity Session
A Decision Clarity Session is a no-obligation conversation where we listen to what you are trying to achieve, what is getting in the way and whether AI, automation, governance, process improvement or decision design is the right next step.
Frequently asked questions
What is Minimum Viable Friction?
Minimum Viable Friction is the smallest amount of intentional pause or resistance needed to improve decision quality without creating unnecessary delay. It is used only where decisions are uncertain, hard to reverse or materially risky.
Is this just another governance process?
No. It should not become another layer of compliance. Minimum Viable Friction is a lightweight decision design tool. It makes assumptions, impacts and judgement visible at the moments where mistakes are hardest to reverse.
Where should friction be added?
Friction belongs at inflection points: major investments, AI use-case approvals, pilot-to-scale decisions, operating model changes, irreversible commitments and decisions that affect people, trust, privacy or risk exposure.
Where should friction be avoided?
Routine, repeatable and low-risk work should remain smooth. Adding friction everywhere creates bureaucracy and weakens the usefulness of the framework.
How does this apply to AI?
AI can generate recommendations and outputs quickly. Minimum Viable Friction helps organisations decide when those outputs need human review, assumption checking, accountability and clear escalation rules.
Can this help with AI governance?
Yes. It turns AI governance into practical decision checkpoints rather than abstract policy. It helps teams know where human judgement needs to remain visible.
How can Changeable help?
Changeable can help map decision points, identify where friction belongs, design lightweight checkpoints, improve workflows and embed Minimum Viable Friction into AI strategy, governance and automation work.
Protect judgement without rebuilding bureaucracy.
Changeable helps organisations design better decision systems for AI, automation and transformation by adding just enough friction where assumptions, impact and accountability need to be visible.