Stop Being the Approval Queue: Give Your Small Team Clear Decision Rights

When every purchase, client reply, design change and deadline adjustment must wait for one person, a small team does not have control. It has a queue.

The pattern often begins with good intentions. A founder wants quality. A manager wants to protect cash. A team lead wants consistent decisions. Soon, however, routine work stops whenever that person is busy. Staff learn to ask before acting, leaders spend their days answering low-risk questions, and nobody is quite sure who owns the result.

The remedy is not vague encouragement to “take more initiative”. It is a practical decision-rights system: clear authority, clear limits, clear escalation triggers and a short record of consequential choices.

First, diagnose the approval bottleneck

An approval bottleneck exists when the flow of work depends on access to one person rather than on the risk and importance of the decision. Typical signals include:

• routine questions repeatedly reaching the founder or most senior manager;
• work sitting idle while someone waits for a reply;
• different people seeking permission for the same kind of decision;
• senior attention being spent on reversible, low-cost choices;
• staff being held responsible for results without having authority to act.

These are system signals, not evidence that the team is lazy or the leader is controlling. Often the underlying problem is that “ownership” has been assigned without specifying the decisions that ownership includes.

Separate input, decision and accountability

Three roles are commonly blurred:

1. Input: the people whose expertise or experience should inform a choice.
2. Decision: the person authorised to choose a course of action.
3. Accountability: the person or group answerable for whether the decision was made properly and reviewed.

Consultation is not collective permission. A decision owner may need advice from finance, operations or a customer-facing colleague, but that does not automatically give each adviser a veto.

This distinction appears in formal governance as well as everyday management. UK Charity Commission guidance says delegated authority should define what a person can decide and when they must report back, while the board retains accountability. It also recommends records detailed enough to show the issues and reasons behind significant decisions. Although that guidance is written for charities, the management principle is useful for any small organisation: delegation works when authority and accountability are both explicit. Read the Charity Commission decision-making guidance.

Build a one-page decision-rights map

Do not begin with every theoretical decision in the business. List the 10 to 15 choices that most often create waiting, confusion or rework. For each one, record five fields:

• Decision: describe the choice in plain language.
• Owner: name one role, not a committee.
• Boundary: state the spending, risk, timing or brand limits.
• Consult: identify anyone whose input is required before deciding.
• Escalate when: write the specific conditions that move the choice upwards.

For example: “The customer-support lead may issue a replacement up to £100 when the order is documented and the customer has not received an earlier replacement. Escalate suspected fraud, safety concerns or repeated claims.” This gives more useful freedom than “Use your judgement” and creates less delay than “Ask the director”.

Delegating decisions to the lowest appropriate level is not merely a speed tactic. It places choices nearer the information. The UK Government’s 2026 spending-control reforms state that decisions should typically sit with a single point of accountability at the lowest appropriate level, while strategic oversight remains higher up. See the Office for Value for Money framework.

Use three lanes instead of one approval chain

A simple way to make the map operational is to divide decisions into three lanes.

Lane 1: decide and act

These are low-risk, reversible and familiar decisions within an agreed boundary. The owner acts and records only what the team genuinely needs to know.

Lane 2: decide, then inform

These choices are still delegated, but they affect another person’s work, a budget forecast or a customer commitment. The owner decides, then shares the result promptly. Notification is not retroactive permission.

Lane 3: consult or escalate

Reserve this lane for novel, high-impact or difficult-to-reverse choices: legal exposure, safety, major spending, a public brand risk, a change to strategy, or an exception outside the agreed boundary.

Current UK digital-assurance guidance similarly argues for greater delivery-team accountability while central oversight concentrates on the largest strategic risks and cross-cutting priorities. Explore the Digital Assurance Playbook.

Keep a decision log that people will actually use

A decision log should reduce repeated debate, not become another reporting burden. For decisions worth remembering, capture:

• the date and decision owner;
• the choice made;
• the main evidence or constraints considered;
• the reason for choosing it;
• who must be informed;
• the review date or condition, if the decision is reversible.

One paragraph is often enough. The aim is to preserve context. Six weeks later, a team should be able to distinguish “we forgot” from “we deliberately chose this because the evidence available then supported it”. Government Digital Service guidance on architectural decisions uses the same logic: record the decision and its context so it can be understood and amended when later evidence changes the situation. See the GDS approach to documenting decisions.

The psychology: make safe autonomy believable

A written map will fail if the social message contradicts it. If people are criticised whenever a delegated choice produces an imperfect result, they will return to permission-seeking. If leaders quietly reverse decisions without explaining why, ownership becomes theatre.

Healthy autonomy needs a fair bargain:

• the team member stays within the stated boundary;
• the leader does not punish reasonable judgement simply because they would have chosen differently;
• mistakes are reviewed for learning, concealment or recklessness—not treated as identical;
• boundaries change when repeated evidence shows they are too narrow or too loose.

This preserves both speed and trust. It also gives leaders a better signal: repeated escalation may indicate insufficient skill, but it may equally indicate an ambiguous rule, missing data or a genuinely risky environment.

Where AI helps—and where it must not decide

AI can help summarise options, extract earlier decisions, draft a short record or flag when a proposal appears to exceed a stated threshold. It can make the information around a decision easier to use.

It should not quietly become the accountable decision-maker. A person must still own the judgement, check the evidence, recognise missing context and understand the consequences. If an AI system cannot explain the business rule it used—or if nobody knows who may override it—the automation has recreated the approval problem in a less visible form.

A 30-minute reset for this week

1. Ask each team member which decision they waited longest for last week.
2. Choose the three recurring decisions with the highest combined delay.
3. Assign one owner to each.
4. Write the boundary and two or three escalation triggers.
5. Decide whether each belongs in “act”, “act and inform” or “escalate”.
6. Run the system for two weeks and review delays, errors and unnecessary escalations.

Do not measure success only by fewer questions. Good questions should still travel. Look for shorter waiting time, clearer ownership, fewer repeated debates and better-quality escalation.

The leadership shift

A leader’s job is not to be the fastest person in the approval queue. It is to design a system in which routine judgement can happen safely without them, important information still reaches the right people, and consequential choices remain accountable.

Start small: one decision, one owner, one boundary and one review date. If the team moves faster without losing control, expand the model. That is how delegation becomes an operating system rather than a motivational slogan.

---

Featured image: conceptual AI illustration created for this article; it is not documentary evidence.