Product Management Is a System of Decisions, Not a List of Ceremonies

The product work I value most has rarely started with a perfectly written brief. It has started with uncertainty: a story that needed a playable form, a website that needed to prove experience rather than merely describe it, or an AI-enabled workflow that needed a real guardrail before it reached someone else.

Across projects such as The Journey, my interactive career world and product delivery work in event, marketplace and AI contexts, I have learned that product management is less about running a fixed set of ceremonies and more about making a sequence of good decisions visible, testable and reversible.

Start with the behaviour you need to change

A backlog can be full and still not explain why a feature deserves to exist. The first question I return to is simple: whose behaviour should become easier, safer or more valuable after this is live?

That changes the conversation. Instead of “we need an onboarding flow”, the work becomes “a first-time visitor can understand the value, decide whether to continue and recover if they leave halfway through.” Instead of “add AI”, it becomes “help a team compress a repeatable piece of analysis without hiding the evidence or removing human accountability.”

Building The Journey made that concrete. It is a personal browser game, but its product questions were familiar: how do people enter, what makes them continue, what happens when a choice is unclear, and how do you make a return visit feel intentional rather than repetitive? The creative format changed; the product discipline did not.

Make the smallest promise you can evaluate

I do not think an MVP means a rough version of every idea. It means the smallest experience that lets you learn whether the central promise is credible.

For a story-led experience, that might be one playable scene with an understandable objective. For a new service or workflow, it might be one clearly bounded handoff. For an AI feature, it might be one task with known inputs, an observable output and a human decision point.

This is where product management earns its keep. It protects a team from solving ten adjacent problems before it has learned whether the first problem matters. It also makes trade-offs explicit: what are we deliberately not building yet, and what evidence would change that decision?

Treat acceptance criteria as a product tool

Acceptance criteria are sometimes dismissed as delivery paperwork. I see them as a way to turn a vague promise into behaviour that a user, designer, engineer and stakeholder can all recognise.

In The Journey, that meant defining more than the happy path. If someone provides an email to unlock an experience, what happens on a repeat visit? What happens if the save fails? Does the experience continue, and is the person told what happened? Those are not edge cases after the fact; they are part of whether the product feels trustworthy.

The same applies to AI-assisted work. “The model produced an answer” is not acceptance criteria. A better definition includes the source material, the response expected from the system, what a person can verify, what should happen when confidence is low and how an error is made visible. The more consequential the workflow, the more important that clarity becomes.

Build the feedback loop before the feature feels finished

Shipping is not the end of product work. It is the moment a belief meets real behaviour.

I try to define the feedback loop before release: what will tell us that the intended outcome occurred, what signals would show confusion or harm, and who will look at those signals? In a public portfolio experience, that can include completion and return behaviour. In a marketplace or payments flow, it can include the points where a person abandons, asks for help or cannot complete a critical action. In an AI workflow, it includes whether people can challenge, correct and trust the output.

The point is not to collect every event. It is to collect enough evidence to make the next decision more intelligent than the last.

Use artefacts to make decisions discussable

Roadmaps, prototypes, acceptance criteria, release checklists and decision logs are valuable because they make thinking inspectable. They give a team something more useful than memory or hierarchy when the trade-off is difficult.

My 3D Career World began as a different kind of product artefact: a way to make a portfolio navigable through projects rather than a list of claims. It forced the same questions I use in delivery work. What should people notice first? Which paths need guidance? What can be discovered rather than explained? And where does visual interest start getting in the way of understanding?

That is a useful reminder for product managers: artefacts do not need to look like a conventional product document to be useful. They need to help people decide.

AI changes the speed of delivery, not the need for judgment

AI has made it much faster to explore interfaces, analyse material, draft alternatives and test implementation ideas. I use that leverage. But speed also makes it easier to ship ambiguity, plausible errors and unexamined assumptions.

The product manager’s role is still to set the problem boundary, protect the user outcome, define what “good” looks like and ensure that accountability remains visible. AI can accelerate the work around those decisions. It cannot make the decisions disappear.

The practical test

When I am unsure whether a piece of work is really product management, I use this test:

• Can we state the user behaviour or outcome we are trying to improve?
• Is there a smallest version that can teach us something meaningful?
• Are the happy path, failure path and recovery path understood?
• Do we know what evidence we will use after release?
• Can the next person understand why we made this trade-off?

If the answer is yes, the work has a chance of becoming a product rather than a collection of output.

The frameworks still matter. So do discovery, delivery rituals and good tools. But the deeper craft is a reliable decision system: one that turns uncertainty into a clear next step, learns in public and leaves people with a product they can trust.