The Fix

Chapter 16: From Pretence to Autonomy

The transition is primarily political rather than structural. This chapter describes what it does to the three populations it threatens most: middle management whose coordination roles disappear, finance teams who lose narrative authority, and product managers whose competence was shaped by the dysfunction the model resolves.

The payments unit was among the first to operate autonomously. Within four months, a pattern emerged: its contract required consuming units to submit fully validated payment requests, which was structurally correct, but the consuming units faced a competing incentive to minimise their own validation costs and let the payments unit reject what it could not process.

A contract arms race followed: the payments unit tightened its schemas, the consuming units added retry logic, and payment failures increased while each unit's metrics looked healthy in isolation and the cross-unit customer journey degraded.

The contract fix was structural: the three units redesigned the contract together, splitting validation responsibilities so that each unit validated what it owned. A shared validation event allowed consumers to pre-check payment-specific fields. The arms race stopped, and the customer experience recovered within a quarter.

The autonomous-unit model had produced both the failure and the fix, but only because the people involved treated the contract as a hypothesis to be tested rather than a boundary to be defended. Scaling the model requires a transitional fund, a CTO-CEO decision pair, and a pull-based relationship between central services and the units they serve. The harder part is the human reality of that scaling, because the transition is primarily political rather than structural.

Moving from a software-dependent corporate to an organisation built around autonomous business units means dismantling structures people depend on, redistributing authority people have earned, and eliminating roles people occupy. It threatens livelihoods and identities, and most attempts fail not because the model is wrong but because the organisation underestimates the political cost of making it real.

Chapters 14 and 15 described the mechanics of the first year: mapping, the first unit, the charter, and the transitional authority. The harder problem is what those mechanics do to people whose competence and identity were formed inside the structure being removed.

Reality-oriented people in organisations where reality is optional are labelled difficult, managed into silence, or filtered out entirely. Lisa, who could map every quarterly priority to its real system constraints and was never consulted, is one of them, as is the operations analyst whose shadow spreadsheet was the only honest record of how the process worked. These individuals never lacked capability; they lacked an operational structure that made their capability relevant.

Inside the unit, the process definition is explicit, the contracts are versioned, and the synthesis runs against artefacts that can receive what these people know. When an engineer says “this transition is not handled,” the claim is verifiable against the definition and the reconciliation output. When an operations analyst identifies an exception pattern, it can be traced against process metrics and either incorporated into the definition or documented as a known deviation.

...

See also: Full contents · Preview chapters · Illusions in the Boardroom