Power and Who Actually Decides

Overview

Structure says who reports to whom. Neither structure nor culture says who decides.

Module 1 introduced the distinction between formal authority and informal power and argued that confusing them is how initiatives die quietly. This topic is the technique: how to map it, how to detect the part that is invisible, and what to do with the result.

The influence and interest grid

Two axes. Influence is how much someone can affect the outcome. Interest is how much they care. Four quadrants, four different relationships — and the useful part is that two of them are commonly mishandled.

Keep satisfied can stop you, isn’t watching — yet

Manage closely the coalition lives here

Monitor

Keep informed your best source of information

Interest → Influence →

The grid, with markers for the roles in a typical initiative. Terracotta is anyone who can stop it; the olive marker in the lower right is the quadrant people neglect and should not.

High influence, high interest — manage closely. Your coalition lives here. These people get individual conversations before anything is presented.

High influence, low interest — keep satisfied. The dangerous quadrant. Someone who can stop you and is not currently paying attention will become interested at the worst possible moment. The move is to give them the one thing they care about, early, so that their first encounter with your work is not an unpleasant surprise.

Low influence, high interest — keep informed. The neglected quadrant, and your best source of information. People who care deeply and cannot decide anything will tell you what is actually happening, who really objected last time, and which previous attempt failed and why. They are also the people who will be using what you build.

Low influence, low interest — monitor. Costs you nothing; check occasionally that they have not moved.

Detecting power that is on no chart

Informal power is invisible by definition, so you need observables. Five that work:

Ask everyone “who else should I talk to?” Names that recur and are not senior are your informal map. Module 1 suggested this; the refinement is to count the recurrences rather than collect the names.
Who gets consulted before a decision is announced? Watch the gap between a decision being made and being communicated — whoever is in that gap has real input.
Whose objection has historically stopped something? The single most predictive question. Ask it about specific past proposals rather than in general.
Who can get a meeting with anyone, quickly? Access is power and it is observable.
Who do people check with informally before committing? Usually revealed by phrases like “let me just run it past…”.

Coalitions, and the order of conversations

No single person approves a cross-cutting change. You need a set, and the set is smaller than everyone — the minimum winning coalition is the smallest group whose combined agreement makes the decision stick.

Finding it is half the work. The other half is sequence:

Talk to the sceptic before the group, not during. An objection raised in a meeting becomes everyone’s doubt, because it is now public and unanswered and other people have watched you be surprised by it. The same objection raised privately a week earlier becomes a design change you present as your own thinking — or a disagreement you have already framed.

Then sequence the rest so that the sceptic hears support from a peer rather than from you. Agreement from someone with no stake in your success is worth more than any argument you make.

Decisions do not happen in the meeting

The structural insight. A meeting named for a decision is usually where a decision is ratified. The deciding happened earlier — in a corridor, a direct message, a one-to-one.

So an architect who only argues in the meeting is arguing after the fact and cannot understand why good reasoning keeps losing. The response is not cynicism; it is pre-socialising: every significant proposal gets an individual conversation with each person in the top-right quadrant before it is presented to anyone collectively.

The cost is real — it is several hours per proposal, it feels like politics, and it is slower than sending a document to a group. It is also the difference between a decision that holds and one that gets revisited. Sections 3 and 4 build the communication content; this is the scheduling.

The map, and the order to work it

Seven people, placed on the grid.

High influence, high interest: Maya Ortiz, who sponsored this and owns the outcome. Priya Shah, whose revenue commitment depends on it. Lena Fischer, who can block outright and is already paying attention because of the personal data.

High influence, lower interest — the dangerous quadrant: Dev Raghunathan, who approved $1.2M a year and will become extremely interested the first time unit economics look wrong. And Tomas Beck, who owns all 140 engineers and whose current interest is mainly in the initiative not disturbing the migration.

Low formal influence, high interest — and high informal power: Ivan Petrov. He appears nowhere in the decision, has no title granting a veto, and can stop this. Grace Adeyemi, who speaks for the 45 agents and whose team’s adoption decides whether any of it works.

The detection evidence for Ivan. Eight years of tenure, owner of the monolith that still runs most of the business, and widely respected. Apply the five observables: his name will recur when you ask who else to talk to, he will be consulted before decisions are announced, and his objection is what quietly ended the 2023 feature’s successor conversations. That last one is the predictive question.

The minimum winning coalition: Maya (sponsor), Dev (money, already committed), Lena (must not block), Tomas (capacity). Four people — and Ivan, who is not in the coalition and can veto it anyway, which is the whole lesson of this module.

The sequence. Ivan first, privately, before anything is presented — because his objection in a leadership meeting becomes everyone’s doubt, and because Module 7’s red-team findings give you something genuine to show him: the automatic regression gate exists specifically to prevent a quiet repeat of 2023. He is the person that argument was built for.

Then Lena, with criteria rather than reassurance, per Module 6. Then Tomas, with the displacement statement from Module 3, because his interest is protecting the migration and a named sacrifice respects that. Then Dev, with the arithmetic including the finding that the company-wide target does not close. Maya and Priya last, because they are already sponsors and need the plan rather than the persuasion.

Grace is not in the coalition and is the highest-value conversation in the list — she will tell you which of the 12,000 monthly tickets are genuinely hard, whether the handoff package is usable, and whether the confirmation volume is survivable. Lower-right quadrant, best source of information, routinely neglected.

Feeds the workbook

This is your stakeholder influence map, and two rubric items land here: it must separate formal authority from informal power and name both, and at least one person’s stated position must be distinguished from their actual interest. Add the sequence — a map without an order of conversations is a diagram rather than a plan.

→ That closes the reading. Now the harder half: what you do about it, given that you cannot change most of it. — Ivory tower or embedded