Outgrace
OutgraceThe Leadership Magazine, home
LeadershipFramework

Which decisions to delegate, and which to keep

Sort recurring decisions by reversibility, exposure and expertise and most delegation arguments end early. This framework is our judgement, not settled research.

Most leaders delegate by mood. A decision lands on the desk, it feels small, and it goes down the line; another feels large, and it stays upstairs. The result is a calendar full of approvals the chief executive could safely skip, and a short list of calls that should never have left the room without a second pair of eyes. This framework sorts recurring decisions on three tests: how easily a wrong call can be undone, how far a wrong call can reach, and who knows enough to make it well. It is our judgement, and we say below where it rests on thin evidence.

Delegation usually goes wrong in one of two directions. Leaders keep decisions that a capable director could make in an afternoon, and they hand away decisions that carry the company's money or reputation with no limit attached. Both errors come from the same habit: deciding by how a choice feels rather than by what it can do if it fails. The tests below replace that feeling with questions that have answers. Together the answers place a decision in one of three bins: delegate freely, delegate with conditions, or keep.

Reversibility: how easily a wrong call can be undone

Start with the undo. Picture the decision going wrong and ask how long it would take someone to notice, and what it would cost to unwind. A discount that can be withdrawn at the end of the month is a different object from a five-year lease. The first is a reversible choice and a strong candidate for delegation, because the damage from a bad call is capped by the time it takes to correct. The second is not a decision to hand away casually, however junior the person who signs it.

Write the exit route down before anything else. For each recurring decision, write one sentence: what we would do, and how long it would take, if this turned out wrong. Where the sentence is short and cheap, delegate with a ceiling. Where it needs a lawyer, a regulator or a year of renegotiation, the decision belongs with someone who can see the whole contract. Most teams find a handful of decisions they treated as minor that have no exit route at all. Those are the ones to fix first, either by writing a route or by moving the decision up a level.

Exposure: how much a wrong call can touch

Exposure measures the reach of a wrong answer: money, people, customers, regulators and the company's name. It is not the same as how hard a decision feels in the meeting. A price change on one product line can be heated and still touch a small share of revenue. A decision about how an incident is reported can look administrative on the agenda and carry most of the risk in the room. Estimate exposure before the debate starts, in the units your board already uses, so the discussion is about the size of the thing and not its tone.

Then set the ceiling in those same units and write it into the delegation schedule. A limit stated as revenue at risk, number of staff affected, or whether an outside body must be told is one a director can check in a minute. A limit stated as a sensible amount is not a limit. In our judgement the most useful ceiling triggers a conversation rather than a veto. Below it, the delegate decides. Above it, the delegate may proceed but must tell a named senior person the same day, so the company learns of the call while it can still be shaped.

Frequency decides whether a decision deserves a rule at all. A decision made twice a year needs a briefing note and a named owner. One made many times a month needs a standing rule, because each call is too small to escalate and the total is large. Count how often each decision recurs before deciding who makes it. Rules for frequent decisions go stale quickly, so put their review on the same date as the ceilings, and let the people applying them report where the rule no longer fits the work.

Expertise: who can make the call well

Expertise asks who knows enough to decide well, which is not always who holds the most senior title. Many poor decisions are not made by people with bad judgement. They are made by people who lack the information that matters, and that information sits two levels down with the engineer, the underwriter or the regional manager. A chief executive who approves every software purchase is, in practice, relying on her memory of the last one and on whatever the vendor has told her.

The test is plain. Ask whether the person you would delegate to could explain the trade-offs to the board, unaided, and answer the two hardest questions a director would ask. If they could, delegate. If they could explain the decision only with a specialist beside them, delegate with that specialist named in the schedule as a required input. If nobody below you can answer the questions, the honest course is to train someone before handing over the authority, not after the first expensive mistake. Expertise can be built, but it is rarely built in the week a decision is due.

Seven questions before you delegate a recurring decision

Answer yes or no for each recurring decision. Three or more no answers mean the decision stays one level up until the gaps are closed.

  • Can we write one sentence describing how this decision would be undone?
  • Can the reversal be completed within a quarter at a cost the board has approved?
  • Is the worst plausible outcome stated in money, people and regulatory terms?
  • Is the delegation ceiling written in units the board already reports?
  • Could the delegate explain the trade-offs to the board without the specialist present?
  • Is there a named person who must be told when the ceiling is crossed?
  • Is there a fixed date on which this limit will be reviewed?

Outgrace checklist

Read the checklist as a set of tests, not a score. A decision that gets yes answers to all seven can be delegated with a light review. A decision that fails the first three questions, the ones about the exit, the cost and the worst case, should not be delegated freely whatever the answers to the others. A single no on the expertise question is a signal to add a required input rather than a reason to keep the decision for good. The most useful output of the exercise is usually the list of questions that could not be answered, because each one names a piece of work someone now owns.

Where it breaks

The framework breaks in three places. The first is the exit route. Teams write that a decision can be reversed without pricing the reversal, and a choice that looks cheap to undo turns out to need a supplier's consent or a regulator's sign-off. The second is exposure measured by feel. A decision with a large tail is sorted as minor because the meeting was short and the numbers were not on the table. The third is authority given without information. A delegate with a ceiling and no access to the figures holds the form of delegation without the substance.

The research base for this sorting is thin, and readers should weigh the framework accordingly. None of the sources we checked for this piece, from the World Bank, the National Bureau of Economic Research and Brookings, sets out a decision-rights framework or reports management research on delegation. The closest item is an NBER training study which says entrepreneurship training may help founders recognise when a venture no longer warrants further investment of time and resources. The study presents this as a possible effect and does not test it as a finding about delegation. It supports the general idea that knowing when to stop is a learnable skill. It does not support the sort order above, which is our judgement.

On Monday, pull the last thirty decisions your leadership team approved, or the last quarter's approval log if you keep one. Sort each on the three tests: can it be undone, how far does a wrong answer reach, and who could explain it to the board unaided. Mark the cases that fail two tests and ask who took them and why. Then choose one recurring category, write its ceiling in the units your board uses, name the person who must be told when the ceiling is crossed, and put a review date in the diary. Send the schedule to your chair before the next board pack goes out.

Newsletter

The Briefing

One email a week: the leaders we profiled, the ideas worth stealing, and nothing else.

Weekly, free, one click to leave. Read our privacy policy.