- The Rule That Breaks Delegation
- What Does “Decision Rights” Actually Mean?
- Why Does Pushing Authority Down Fail?
- Part Two: Assigning the Decision
- Four Questions That Name the Person
- What About Decisions Nobody Owns?
- About the Stagnation Assassin
Pushing a decision down to the person closest to the work makes it worse about as often as it makes it better. That is not an argument against delegation. It is an argument against the version of delegation that moves authority and leaves everything else where it was.
The objection arrives immediately, so let me take it head on. Yes, the person closest to the work knows the most about the work. They frequently do not know the most about the decision, because a decision involves consequences outside the work, and proximity to the task does not confer visibility into what the task connects to. A scheduler optimizing a sequence knows the sequence better than anyone. Whether the sequence should protect a customer who is currently in a sourcing review is not a scheduling question, and the scheduler may have no way to know it is happening.
The Rule That Breaks Delegation
Authority without information produces one of two outcomes, and both look like the delegation failed. Either the person decides confidently with a partial picture and gets it wrong in a way nobody predicted, or they sense the gap and escalate for safety, which adds a step to the process you were trying to shorten.
The second outcome is by far the more common and the more expensive, because it is invisible. Nothing goes wrong. The decision just takes longer than it did before delegation, and when someone notices, the conclusion drawn is that the team was not ready for the responsibility. The team was ready. The design was incomplete.
Authority without information does not produce bad decisions. It produces cautious escalation, which is slower than what you had, and it gets blamed on the people rather than the design.
So the rule is simple and rarely applied: authority moves only together with the information and the context that decision requires. If you cannot move both, do not move either, because moving one alone is worse than leaving it.
What Does “Decision Rights” Actually Mean?
Decision rights name three roles for a specific class of decision: who decides, who must be consulted before deciding, and who is informed after. Three roles, written down, for decisions that recur.
The part people skip is “for a specific class of decision.” Decision rights assigned to a person or a job title are useless, because nobody makes decisions in general. They make pricing exceptions, expedite approvals, overtime calls, and capital releases, and the right answer for one is frequently the wrong answer for another. A scheduler might hold full authority over sequencing, be a consulted party on expedites, and be merely informed about pricing. That is three different positions for the same person, and collapsing them into a single seniority level is what produces the mess.
The other part people skip is that consulted is not the same as agreeing. A consulted party contributes information and does not hold a veto. Where that distinction is not made explicit, consultation quietly becomes consensus, and consensus is the mechanism by which a decision with a named owner still takes eleven weeks.
Why Does Pushing Authority Down Fail?
Three specific failures, and each has a signature you can spot inside a month.
The first is the information gap already described. Signature: decisions that used to take a week now take three, with no increase in quality and no complaints from anyone.
The second is unbounded authority. Someone is told they now own a class of decision without being told the limits, so they are effectively being asked to guess where their authority stops. Rational people guess conservatively. Signature: the delegated authority is used for the easy cases and escalated for exactly the cases where a decision was actually needed.
The third is retained veto. Authority is announced as delegated and then a decision gets reversed from above, once, publicly. That single reversal ends the delegation regardless of what was announced afterward, because everyone now knows the real rule. Signature: a documented delegation that nobody uses, and a leadership team puzzled about why.
One public reversal ends a delegation permanently. Everyone updates on the reversal, not the announcement, and no amount of restating the policy undoes it.
Notice that none of the three is a failure of the person receiving the authority. All three are design failures, which is useful, because design failures are fixable and character assessments are not.
Part Two: Assigning the Decision
From here the work is mechanical. Take the decisions that recur, run each through four questions, and the decider is usually named within a minute. The questions are ordered deliberately, because the first one eliminates most of the ambiguity before the harder ones arrive.
Four Questions That Name the Person
Ask them in this order for each recurring decision, and stop as soon as the answer narrows to one person.
One: who holds the consequence? Not who does the work, and not who has the most expertise. Who carries the result if it goes badly. That person is the default decider, and the burden of proof sits with anyone arguing for someone else.
Two: what information does the decision actually require, and who has it? If the consequence-holder lacks the information, you have two choices and only two. Move the information to them, which is usually possible and usually cheaper than it looks, or move the decision to whoever has the information and give them the consequence too. What you cannot do is leave the split in place and hope.
Three: how reversible is it? Reversible decisions go to the person closest to the work, because the cost of being wrong is bounded and the cost of being slow is not. Irreversible decisions go up, and the number of genuinely irreversible operating decisions is far smaller than most organizations behave as though it is.
Four: who has to live with it that we have not mentioned? This is the consulted list, and it should be short. Two names is normal. Five names means you are building consensus with extra steps.
| Decision class | Decides | Consulted | Informed |
| Production sequence within the week | Scheduler | Plant supervision | Customer service |
| Expedite at the cost of another order | Plant manager | Sales owner of both accounts | Scheduler, customer service |
| Overtime authorization | Plant manager, to a stated dollar limit | None below the limit | Finance, weekly |
| Price exception on a standard line | Sales leader, to a stated discount limit | Finance above the limit | Pricing owner |
| Capital release under a stated threshold | GM | Finance, engineering | Leadership team |
Notice the stated limits. Three of the five rows carry a number, and the number is the thing that makes the delegation usable. Authority with a named boundary gets exercised. Authority without one gets escalated, because the safest reading of an unbounded grant is a narrow one.
What About Decisions Nobody Owns?
They are the most expensive decisions in the business and they are invisible, because an unowned decision does not appear anywhere as a delayed item. It appears as a recurring topic that keeps coming up and never resolves.
Find them by asking three senior people who decides a given class of decision. Three different answers means nobody does. The follow-up that confirms it: ask who made the last one. The pause before the answer is the diagnostic, and the answer, when it comes, is usually a description of a process rather than a name.
Assign them immediately and imperfectly. A decision assigned to the roughly right person today beats a decision assigned to exactly the right person after a two-month design exercise, because the assignment can be corrected in five minutes and the two months cannot be recovered. The unowned decisions have usually been unowned for years, and the cost of that has already dwarfed the cost of getting the assignment slightly wrong.
An unowned decision never shows up as a delay. It shows up as a topic that keeps coming back and never resolves, which is why it survives for years.
Where these assignments get exercised week to week is a structural matter, and it belongs in the operating cadence rather than in a policy document. The broader question of where decisions are waiting in the first place, of which unclear authority is only one of five answers, is covered in the five places decisions get stuck.
One limitation worth naming: written decision rights work well for decisions that recur and badly for decisions that are genuinely novel, which by definition have no class to belong to. Novel decisions still need a human judgment about who should hold them, made in the moment. Any organization that responds to this article by attempting to document its way out of that judgment will produce a large document and the same ambiguity it started with.
About the Stagnation Assassin
Todd Hagopian is a Fortune 500 transformation executive who has generated $3B in shareholder value across Berkshire Hathaway, Illinois Tool Works, Whirlpool, and JBT Marel, where he serves as Global VP of Product Strategy and Marketing. Known as The Stagnation Assassin, he is the author of two published books: The Unfair Advantage: Weaponizing the Hypomanic Toolbox and Stagnation Assassin: The Anti-Consultant Manifesto. His blog is published in 15+ languages and read by operators worldwide. Bring him to your stage via his speaking page or connect with him on LinkedIn.
For leadership teams. Decision rights sits inside the wider turnaround material I speak on. Formats and booking are on the speaking page.