Authority levels from recommend to veto
A decision rights matrix assigns each decision type to an authority level. Recommend (propose an option, but someone else decides). Decide (you choose, others execute). Approve (you must sign off before the decision sticks). Inform (you are told the outcome, no input). A junior engineer might have Recommend authority on architectural choices, the tech lead has Decide authority, and the VP may have Approve. Different organizations distribute authority differently, but the discipline is the same: clarity about who holds which power prevents blocked decisions and second-guessing.
The key distinction is between Decide and Approve. A person with Decide authority owns the outcome; they are responsible if it goes wrong. Approve authority is a veto: you can block a decision but are not responsible if it succeeds. This matters for accountability. Too many Approve layers create paralysis. Too few Recommend and Decide authorities underutilize the expertise in the organization.
Scaling through distributed authority
As companies grow, centralized decision-making becomes the bottleneck. If every decision above a certain size needs CEO approval, the CEO becomes a scheduling constraint. Mature organizations push Decide authority down to team leads and individual contributors. The CEO keeps Approve authority on rare, high-stakes decisions. This creates speed: decisions move forward because authority is distributed; the CEO intervenes only when necessary.
The trap is inconsistent distribution. If the matrix is unclear or changes per person, different employees have different power levels despite similar titles. This breeds resentment and prevents delegation. Clear matrices make scaling possible: a new team can be onboarded into the decision framework without renegotiating every decision. Authority becomes an attribute of the role, not the person.