Segmentation

Project management technologies evolve — and with them, we evolve as well. Nowhere is this more visible than in IT.

Not so long ago, companies were forced to think systemically. Product updates were released once a year, sometimes even less frequently. New versions often brought not only functional improvements, but also deep architectural changes.

Today, we think in increments.

Updates are shipped at a pace that would once have seemed absurd. Scrum, DevOps, and the overall tempo of modern work have shaped this mindset. There is simply no time to rethink architecture anymore: it does not fit into an increment, and it cannot be rebuilt within a sprint. As a result, products increasingly resemble the Ship of Theseus. After countless small changes, not a single “old plank” remains.

Formally, the ship is entirely new — every component has been replaced. In essence, however, it has changed far less than it appears.

We have learned to think in backlogs rather than architectures. This way of thinking inevitably shapes our products. Products shape our digital environment — and that environment, in turn, reshapes us. Individuality gradually adapts to popular methodologies, producing increasingly uniform project teams.

And yet, intuitively, we know the truth should be the opposite: methodologies should adapt to companies, not force companies to reshape themselves around a method.

Real success comes through individuality.

It is hard to win when you are competing against teams structured exactly like your own. By repeating someone else’s experience, you walk a road that has already been traveled — where all the prizes have long been claimed. No one has ever become a millionaire by reading a book titled "How to Become a Millionaire".

In the first article of this series, I showed what project management methodologies have in common. In the second, how they can be combined. Together, they outline a path by which a team can form its own project technology — one that reflects its individuality and creates genuine, useful uniqueness.

This is the trajectory where a team can remain itself.

But freedom always comes with responsibility.

It requires understanding how a chosen approach actually works — and why it will work for you. This article is dedicated to that question. It is an attempt to show how segmentation inevitably emerges within any project, why it is not a trick or a managerial gimmick, but something rooted in the very nature of projects themselves.

A Simple Project Model

Let us begin with a deliberately simplified project model.

All effort within a project can be conditionally divided into two groups.

• Production activities are the protagonist actions. They directly move the project toward its goal and advance its “plot”.
• Management activities ensure that production efforts remain controlled and actually lead to the desired result.

A simple metaphor works well here. Production is a fast horse racing toward the goal. Management is the rider guiding it. Separately, each is far less effective. The rider moves slower than the horse, but knows the route — knows when to stop, when to turn, and when the horse needs care.

Now consider a single executing activity.

At any moment, it exists in a certain state, defined by three things:

• how closely the current result matches the goal,
• the quality of execution,
• and how far the activity has progressed relative to expected time.

At the start, the activity is far from the goal. Ideally, by completion, it should fully reach it — with the required quality and within expected time.

During execution, people constantly make local decisions. Each slightly alters the direction of movement toward the result. In an ideal world, these decisions form a smooth, predictable trajectory. In reality, errors, approximations, and assumptions are unavoidable. The actual trajectory begins to deviate from the optimal path.

This deviation accumulates and reflects a loss of controllability.

Crucially, once the first deviation occurs, returning to the optimal trajectory becomes increasingly difficult. Errors trigger further decisions. Even if those decisions are reasonable in the local context, they tend to reinforce the initial drift.

This is where management activities become essential.

Their role is to periodically assess the current direction and correct it, reducing accumulated deviation. Such interventions are neither instantaneous nor perfect: they take time and introduce their own inaccuracies. Viewed over time, the total error usually continues to grow — just more slowly.

Eventually, a point is reached where continuing without pause only worsens the situation. Deviation grows faster than management can compensate. At this moment, the activity must be slowed down or stopped, allowing management to “catch up” and restore control.

In real projects, many activities execute simultaneously. Deviations accumulate across all of them. When overall project uncertainty grows faster than management can reduce it, the project must be stopped — partially or entirely.

These forced pauses are the essence of segmentation.

Returning to the horse analogy: when a rider loses direction as dusk approaches, the instinctive response is to spur the horse and move faster. Yet this is often the fatal mistake. Speed in a random direction does not bring you closer to the goal — it may carry you so far away that the goal becomes unreachable.

In projects, direction matters more than speed.

Even slow, step-by-step progress can reach the goal, provided movement remains controlled. “Step by step” is the leitmotif of segmentation. Its purpose is not to slow progress, but to preserve control over direction.

Once the mechanism is understood, a natural desire emerges: to understand how it works in detail — and how to do it better.

Production Disorientation

The root cause of all the issues described above is deviation of production activities from the optimal direction. Let us examine how this deviation arises.

Multiple scientific disciplines point to the same fact: each person’s mental model of the world is subjective. We perceive the same events differently, interpret identical formulations in different ways, and assign different meanings to the same concepts. Under these conditions, it is entirely natural for people working on the same activity to understand its goal differently.

This is the first source of deviation — before any work begins, at the level of interpretation.

How can this divergence be reduced at such an early stage?

Consider shooting at a target. Individual shots inevitably deviate from the center. Yet the midpoint between two shots is usually closer to the center than either shot alone. This is a general property of random error: when deviations are random, averaging reduces dispersion.

Formally, if individual understanding errors have a characteristic spread, then combining several independent viewpoints reduces the spread of the average proportionally to the square root of their number. With two people, error decreases by roughly √2 — nearly one third. With four, it halves.

Applied to projects, this means that teamwork reduces goal distortion even before formal control mechanisms appear. Simply forming a team already lowers the risk of misalignment, especially at early stages.

The same principle explains the simplest management intervention. A person performing a corrective role temporarily joins the team. Even without formal procedures, their presence statistically reduces error by adding another viewpoint and cross-checking interpretations.

There are many other ways to reduce subjective error: knowledge bases, metrics, checklists, milestones, reviews, tests, training. Fundamentally, however, they are all forms of delayed collective work — mechanisms for capturing and reusing accumulated experience.

Organizing Work

Psychological and sociological studies consistently show that people tend to engage more intensely as deadlines approach. Temporal Motivation Theory describes this effect well: the motivational “weight” of a task increases as remaining time shrinks, even if the task’s objective value stays the same.

Deadlines sharpen focus.

This trait of human behavior is remarkably stable and difficult to overcome directly. In project management, it is therefore wiser to work with it rather than against it — structuring boundaries and segments that harness rising motivation without sacrificing control.

An elegant solution comes from the Theory of Constraints. Experts often estimate task duration as T₉₀ — the time within which a task will be completed with about 90% probability. Such estimates include significant buffer, so tasks often finish earlier.

A more practical approach is to define T₅₀ — the time within which the task completes with 50% probability — and use it as the target for executors. A dedicated buffer of size T₉₀ − T₅₀ is placed at the end.

Planning reliability remains intact: total duration is still T₉₀. But executors work toward a more realistic and motivating target, reducing procrastination and improving controllability.

In effect, one task is split into two activities: a production activity and a methodological one. This can be seen as a trick, but a more straightforward solution would be to decompose the task into several independent ones. Multiple deadlines create multiple focus points, increasing engagement and momentum.

At the project level, this same principle applies. Such structuring can be viewed as organizational segmentation — a form of artificial segmentation driven not by product structure, but by management logic.

Execution Control

Almost any competent professional spends part of their effort on self-control: checking results and fixing mistakes. On average, this consumes 5–20% of working time, often much more for beginners.

Self-control is effectively a hidden management activity. While it stays within reasonable bounds, it is beneficial: it speeds work, builds skill, and preserves flow. But once it exceeds roughly a third of total effort, it becomes a source of loss.

The main cost is not time itself, but lost opportunity. Experience gained through individual self-control accumulates poorly and scales badly. It remains local and quickly depreciates.

By contrast, a dedicated testing or review session performed by another specialist typically produces documented results, formalized conclusions, and reusable knowledge. This increases organizational maturity, not just individual skill.

Self-control must therefore be managed.

Teams should cultivate tolerance for ongoing errors and a shared understanding that mistakes are a normal part of work. What matters is not the presence of errors, but situations where correcting minor deviations consumes more effort than they objectively deserve.

Naturally, this reasoning does not apply to systems with high cost of failure — critical infrastructure, medical systems, or industrial production.

Task Duration

During execution, people constantly make local decisions. Each slightly shifts the state of the activity. The number of truly correct decisions is usually much smaller than the number of possible mistakes, which makes deviation from the optimal path almost inevitable.

Over time, these deviations accumulate. The longer an activity lasts, the more pronounced the effect becomes. For long-running tasks, accumulated error becomes a risk factor in its own right — even if individual decisions seem reasonable.

A tunnel-vision effect also emerges. Prolonged focus on a narrow task area erodes holistic understanding. Teams lose sight of interdependencies, fall out of sync with parallel work, and begin optimizing local goals at the expense of the whole.

The simplest response is artificial decomposition of a long task into smaller activities. But this raises an important question: by what criterion should the split be made if the task has no natural internal structure?

Agile methods propose an obvious answer: time-based slicing into sprints. This is effective and widely used, but it remains a linear approximation of a much deeper problem.

The art of segmentation lies in knowing when to pause — not too often, but without losing control. Not stopping the entire project, but intervening precisely where needed, and correcting course surgically.

A detailed exploration of segmentation mechanics lies beyond the scope of this article and deserves a dedicated discussion.

feedback@shortki.com user support