The Gap
This is Part 3 of a series exploring product management as a system — not as a set of practices.
What Happens When the Textbook Meets the Reality
There is a model of product management that the industry has agreed on as the ideal. It lives most completely inside the companies that built it — Google, Meta, Amazon, Apple, Netflix. The PM owns the problem space. Decisions are made with data. Experiments validate assumptions. Teams are empowered to solve problems rather than just execute features. This is the textbook version — real, coherent, and built on specific enabling conditions that most companies do not have.
And then there is the reality. Most PMs work in one of several distinct operating modes — shaped not by choice but by the organisation’s structure, the distribution of decision authority, and the conditions available to work within. The Operator manages delivery in a structured programme environment. The Negotiator brokers alignment across stakeholders with competing claims. The Experimenter runs learning cycles in data-rich consumer contexts. The Architect manages shared systems that other teams build on. The Pioneer reduces uncertainty in unmapped territory. Each mode has its own logic, its own definition of success, its own constraints.
This article is about what happens when the textbook meets the reality. When a PM in one of these modes tries to operate like the ideal — or is expected to, by their organisation, their manager, or themselves.
The Gap Is Not Obvious
The gap between the ideal and the operating mode reality does not announce itself clearly. It does not arrive as a single moment of recognition. It accumulates.
It accumulates in practices that lose their function. The experiment still gets run — but it cannot fail, because the change was already committed. The OKR still gets written — but it is a feature list dressed in outcome language. The PRD still gets produced — but it is written to secure approval, not to discover what should be built. The form persists. The function is gone.
It accumulates in ownership that does not hold. The job description says the PM owns the problem space. The organisational structure says the decisions were made upstream. The PM is accountable for outcomes they cannot control, using tools that were designed for a different kind of authority than the one they actually have.
And it accumulates in something harder to name — a growing sense that the problem is you. That you are not data-driven enough. Not strategic enough. Not influential enough. That other PMs in other companies are doing the real version of this work, and you are doing something lesser.
That sense is the most consequential part of the gap. Because it is not accurate. The problem is not the PM. It is the distance between the conditions the model requires and the conditions the context provides.
What the Gap Looks Like From Inside Each Mode
The gap does not look the same from every position. Its shape depends on which operating mode you are in — because each mode has its own version of what the ideal demands and what the reality provides.
In the Operator mode, the FAANG ideal says: own the outcome, drive the roadmap, make decisions based on evidence. The Operator’s reality is that decisions were made before they arrived. The roadmap came from above. The data, where it exists, is operational — delivery timelines, ticket volumes, sprint velocity — not the user behaviour signals the ideal assumes. The gap here produces a specific kind of performance: the PM writes outcome-oriented OKRs that are, underneath, a list of features they were already told to build. They run experiments on changes that were already committed. They produce the artefacts of empowered product management while operating inside a delivery engine. The form is present. The function is absent.
In the Negotiator mode, the ideal says: use data to resolve disagreement, align stakeholders around user value, let evidence drive prioritisation. The Negotiator’s reality is that data does not settle disagreements in a veto-rich environment. Stakeholders have legitimately different goals — revenue, risk, retention, relationship — that cannot be collapsed into a single metric. The gap here produces a different performance: the PM learns to use data as a negotiation tool rather than an arbiter. The PRD becomes a document for manufacturing consent. The OKR becomes a boundary object that different stakeholders can interpret differently. The artefacts travel. Their function — to make decisions based on evidence — does not.
In the Experimenter mode, the gap is subtler because the conditions are closest to the ideal. The Experimenter has access to data, to experimentation infrastructure, to measurable user behaviour. But the conditions are rarely as complete as the ideal assumes. Traffic may be insufficient for statistical validity. Tests may be designed to validate rather than to learn. The organisation may treat “statistically significant” as “true” without examining effect size or practical significance. The gap here produces a specific failure: the PM runs experiments designed to confirm rather than to learn. Shipping velocity gets mistaken for learning velocity. The ritual is present — hypothesis, test, result. The learning is not. The organisation believes it is operating like the textbook. The decisions are no better than before.
In the Architect mode, the ideal says: define the customer experience, ship iteratively, let metrics guide trade-offs. The Architect’s reality is that the system is tightly coupled, the feedback loops are long, and the outcomes that matter — developer productivity, system resilience, scalability — are mediated by many downstream teams and releases before they appear in any metric. The gap here produces a particular frustration: the PM produces strategy documents and technical roadmaps that describe a future state the organisation cannot fund or execute. The vision is clear. The path from here to there is blocked by legacy constraints, competing priorities, and the simple fact that platform work has no budget separate from feature delivery.
In the Pioneer mode, the ideal says: run experiments, observe user behaviour, let evidence reveal the right product shape. The Pioneer’s reality is that there are no users yet, no behaviour to observe, no experiments that can produce statistically valid results. The gap here produces false precision: the PM measures what can be measured — prototype clicks, survey responses, waitlist signups — and treats those signals as validation. The metrics exist. What they measure is not what matters. The organisation believes progress is happening. The signal is noise.
Why the Gap Persists
Each of these distortions is predictable. Each follows directly from the mismatch between what the ideal assumes and what the operating mode provides. And yet the gap persists — in organisation after organisation, role after role.
The reason is structural. The gap is invisible at the point where it matters most: before anyone starts the job.
Job descriptions are written in FAANG language regardless of the actual operating mode. Not because companies are dishonest, but because FAANG language is the professional standard — what a modern PM role is supposed to sound like. The description promises outcome ownership, data-driven decisions, empowered teams. The structure delivers something different. Neither party has lied. But the expectation has been set against a standard the context will not support.
The interview reinforces it. Candidates demonstrate FAANG-shaped thinking. Companies evaluate FAANG-shaped criteria. Both sides perform the ideal — not cynically, but because it is the only shared vocabulary available. The mismatch is not visible in the room. It becomes visible six months later, when the PM tries to exercise the authority the job description implied they had.
By then, the box has been drawn. Expectations have formed on both sides. The PM has begun adapting to the actual operating mode — not because they chose to, but because adaptation is how you survive in a system that does not provide the conditions the stated model requires. The box is not imposed all at once. It forms gradually, through accumulated accommodations, each one individually reasonable, collectively constraining.
The Organisation Copies the Form
The PM’s experience of the gap is one side of it. The organisation’s contribution is the other.
Most companies that describe themselves as product-led, data-driven, or empowered have adopted the vocabulary of the FAANG model without building the conditions that make it functional. OKRs get introduced as a goal-setting tool — but goal-setting authority stays centralised, so OKRs become a reporting layer rather than a delegation mechanism. Teams are told they are empowered — but the decisions about what to build, when to build it, and what counts as success are still made upstream. “Data-driven” becomes a cultural aspiration — but the data infrastructure cannot support the decisions being made, so dashboards decorate conclusions that were reached by other means.
The observable difference between a company that has genuinely adopted the model and one that has adopted its surface forms is not the presence of the artefacts. Both have OKRs. Both use product discovery language. Both describe their PMs as owners. The difference is whether those practices actually change what the organisation can decide, and how fast it learns.
Where the conditions are present, a PRD is a decision instrument — it clarifies what should be built and why, and the organisation can act on it. Where the conditions are absent, a PRD is an approval instrument — it documents what has already been decided, in language that sounds like product thinking. The artefact looks identical. Its function is opposite.
This is why the gap persists even in companies that believe they have closed it. The vocabulary of the model travels easily. The enabling conditions — delegated decision rights, fast feedback loops, reliable data, loose system coupling — do not travel at all. They have to be built. And building them is slower, harder, and less visible than adopting the language.
The Moves That Go Unseen
The same role, the same structure, the same constraints — and yet the degrees of freedom available within any operating mode are not zero. The Operator who has been in a delivery structure for two years has developed real product judgment — pattern recognition around which dependencies are silently at risk, which stakeholder needs alignment before the meeting not during it, which scope trade-off will cause problems three sprints later. That judgment is a form of product leadership. It operates at a different altitude than the textbook describes, but it is not absent.
The gap, however, makes it hard to see. When the only available reference point is the full FAANG ideal, the current context looks like an almost total absence of the thing the ideal describes. The distance feels total. So the available degrees of freedom go unrecognised — not because they do not exist, but because the reference model is too far away for them to register as meaningful.
This is the third structural effect of the gap. The first is that practices lose their function when applied without the enabling conditions. The second is that the organisation reproduces the vocabulary without the system. The third is that the ideal, by being so complete and so distant, makes the available space within constrained modes structurally invisible — not to an outside observer, but to the PM operating inside it.
What the Gap Actually Is
The gap is not a failure of execution. It is not a failure of ambition or capability. It is the predictable result of a model being applied to conditions it was never designed for, without any shared vocabulary for naming what is happening.
The practices lose their function not because PMs apply them badly, but because the enabling conditions they require are absent. The ownership does not hold not because PMs lack authority, but because the organisational structure was never designed to give it. The sense that the problem is you is not an accurate diagnosis — it is what happens when a structural constraint has no name and no shared language, so it gets interpreted as a personal failing.
The gap does not announce itself. It accumulates — in practices that lose their function, in ownership that does not hold, and in a growing sense that the problem is you.
It is not.