Distributed Decisions: The Advice Process, Self-Management, and Living Purpose
Most organizations pick between two bad options: every decision funnels up to one person, or nobody owns anything and decisions drift. There's a well-worn middle path used in self-managing organizations, and it starts by dropping a single assumption — that a decision needs everyone's agreement to be legitimate.
The advice process: authority to the doer, wisdom to the system
Anyone can make almost any decision, on one condition: they first seek advice from the people it meaningfully affects and the people with real expertise in it. Consensus is not required — hearing them out is. The decision-maker weighs what they heard and decides; they don't need every advisor's blessing, only a genuine reason for whichever piece of advice they didn't follow.
High-risk or regulated decisions still deserve tighter, pre-agreed limits — the advice process isn't a way to skip governance on the decisions that genuinely need it. It's meant for the much larger category of everyday decisions that currently wait in a queue for someone senior, purely out of habit.
Picture a product designer who wants to change the onboarding flow. Instead of writing a proposal and waiting for a director's sign-off, they talk to two support agents who field onboarding complaints and one engineer who'll build it, hear a real concern about a specific edge case, adjust the design to handle it, then ship the change and post a short note explaining what changed and why — including the one thing they didn't change despite the advice. No approval meeting happened. Three people who actually knew something relevant were heard, and everyone else can see the decision and its reasoning after the fact.
Three conditions before you remove a rule
1 · Healthy baseline
The principles of a good decision, ethics, and risk limits are commonly understood.
2 · Visible information
The minimum data that would show an early deviation is shared transparently.
3 · A conversation forum
A regular space exists where that data gets interpreted and turned into corrective action.
A finance team once removed a manager pre-approval step for expenses under a set threshold — but only after building a weekly dashboard that flagged unusual spending within days instead of the quarterly audit that used to catch it three months later, and a standing quick review of that dashboard every Friday. The rule went away; the actual control got faster, not weaker.
What self-management actually requires
Required infrastructure
- • Clear purpose and roles
- • A distributed decision mechanism
- • Transparent information flow
- • A way to resolve conflict
- • Coaching and support
Common misreading
- • "There's no structure or coordination"
- • "Every decision needs unanimous vote"
- • "Everyone has equal say on everything"
- • "Direction emerges on its own"
A team that hears "self-managing" and drops its planning rituals without replacing any of the left column isn't self-managing — it's unmanaged, and it usually finds that out the hard way within a quarter.
Purpose as a compass, not a fixed destination
Predict and control
Turn the future into a detailed plan, treat deviation as failure, push information up, push execution down.
Sense and respond
Hold the long-term direction and purpose steady; continuously update the near-term plan with new signals.
This isn't a lack of direction — purpose isn't a fixed destination, it's the shared compass used to pick the next best step. In practice it flows down and gets tested with evidence flowing back up:
Take a company whose purpose is helping small retailers compete online. A strategic outcome might be reducing the time it takes a new merchant to make their first sale. The bet they're testing this quarter is that a guided setup wizard, not more documentation, is what's actually blocking merchants. One team's Product Goal narrows that to "merchants can list their first ten products without contacting support," and this sprint's Sprint Goal narrows it further to proving that with one specific product category. Not every task needs to trace back to the purpose explicitly — but everyone should be able to say which bet their current sprint serves, and real usage data flows back up to tell leadership whether that bet is paying off or needs to change.
Now practice it
You own a real decision. Gather genuine advice without seeking consensus, decide, and share your reasoning transparently — with AI rubric feedback.
Open the PracticeLab →Frequently asked questions
Isn't the advice process just consensus with extra steps?
The opposite — consensus requires everyone to agree, which is exactly what slows decisions down or waters them into mush. The advice process keeps one clear owner; it only requires that owner to genuinely listen before deciding.
How do you know when it's safe to remove an approval step?
Only once all three conditions hold together: the team shares a clear enough sense of what a good decision looks like, the data that would reveal a bad one is visible in near real time, and there's a regular forum where that data actually gets discussed.
Doesn't 'sense and respond' mean giving up on planning?
No — it means treating a plan as a current best guess instead of a fixed commitment. The long-term purpose stays stable; the near-term plan updates continuously as new signals arrive.