The customer wanted to change the annual mileage on her renewal from 6,000 to 8,000. The box on the insurer's page was grey. Technology estimated a few days. Nine months after the request entered the portfolio, a programme manager was taking the correction to a steering committee.
A company groups people to do useful work. Customers judge the products and services; shareholders judge the economics. Neither pays for the boxes.
Inside are careers, budgets, systems and rights. Corporate politics begins when protecting them takes precedence over the customer or the economics. Good design puts personal incentives in contact with a complete outcome.
Alignment sounds virtuous because people should understand one another. A company needing persistent alignment for ordinary change is usually admitting something else: it has divided one outcome among partial owners. Each can delay the whole; nobody can complete it. Committees and bridge roles make the stalemate habitable. Coordinating the result starts to count as leadership.
Throughout this book, “alignment” means that corporate work: the recurring effort required to coordinate a result across divided authority. The subject is organisational design in software-dependent companies, not the alignment of an artificial-intelligence system.
The insurer in this book does not exist. No single mileage request passed through every scene exactly as described. The company, characters, chronology and results are composites assembled from recurring situations encountered during more than twenty-five years of work with organisational and software systems. The dialogue is written, not transcribed, and the numbers are invented to make the mechanism testable rather than to report a disguised company.
Keeping one request alive exposes the gap between an account and a mechanism. A green box can be locally true while no customer has used the change. Every scene asks whether somebody closes that gap, is processed by it or earns a living making it presentable.
Felix produces the account. Marianne first helps sell it and later joins the board that must inspect it. Lisa tries to close the gap from inside; Martin has built a career in it; Elliot knows the mechanism that the account edits away; Nadia sees what happens to the customer. Other people remain named by function. That is how the company first knows them.
Within the fiction, Felix reconstructs the story from decision records, emails, release evidence and later interviews with Lisa, Martin, the control holder and Elliot. Where memory and evidence disagree, the book keeps the disagreement.
The proposed correction puts ordinary decision rights inside a complete process boundary while retaining independent control. The first unit works partly because Lisa is unusually capable. The next test is whether another unit can complete ordinary decisions in a less forgiving process without her.
The loss at stake is competitive. While one company spends a quarter assembling permission, another can learn from the same customer problem each week.
Leaders can follow the story and finish with What to Do on Monday. Practitioners can use the Part essays and the companion Technical Field Guide to inspect the mechanism in full. Anyone who loses a term can treat the Glossary as a map.
The authority to move rights usually sits with a chief executive, chief operating officer or board. The evidence often starts with an engineer who can trace the state, an operator who sees the exception or a control specialist who can state the real constraint. One group can change the allocation. The other can produce an account it can no longer avoid.
“Show me the incentive and I'll show you the outcome.”
– Charlie Munger