Glossary

Manage by Exception

Manage by exception is a control principle, formalised in PRINCE2, where a manager agrees tolerances — the limits within which work can run without sign-off — and is only alerted when a forecast breaches them. It lets senior people delegate the routine and reserve their attention for the genuine exceptions.

Manage by exception is a control principle in which a manager defines the boundaries within which work can run on its own, then steps back and is only alerted when a forecast shows those boundaries are about to be crossed. It is one of the seven principles of PRINCE2, the project-management method governed by Axelos (now part of PeopleCert), where it is paired with the idea of delegated tolerances — explicit limits across time, cost, scope, quality, risk and benefit.¹²

The mechanism is simple. A manager agrees the tolerance with the person running the work, who then has the authority to manage everything inside it without asking permission. The moment the work is forecast to breach a limit — not when it already has — an exception is raised and the decision moves up a level.

Why it matters

Manage by exception matters because attention is the scarcest resource a leader has, and treating every item as equally urgent wastes it. By agreeing tolerances up front, a manager delegates the routine and keeps their focus for the genuine exceptions — the things that have actually moved outside the safe envelope. This is also what makes a RAG status meaningful: a clear tolerance is what lets "amber" mean something specific rather than a vague mood. Without defined limits, escalation becomes a matter of nerve and noise, and the loudest problem wins rather than the most important one.

Escalate before the breach, not after

The principle's real value lies in its timing. The Control Standard makes the same point: escalation should be a low-drama, rehearsed move that travels before surprise becomes the management system — triggered the moment a date, margin, risk or quality forecast moves beyond the agreed tolerance, while there are still options to protect the outcome. A RAID log helps here, by keeping the risks and dependencies that might breach a tolerance visible and owned. The discipline is the same one that keeps any work under control: agree what "within limits" means, and make crossing it an event rather than a discovery.

Axelos / PeopleCert, Managing Successful Projects with PRINCE2 — the "manage by exception" principle and the six tolerance areas. See the official PRINCE2 overview at prince2.com. ↩
PeopleCert, PRINCE2 — the world's most widely adopted project management method — peoplecert.org. ↩

Frequently asked

What are tolerances in manage by exception?

Tolerances are the agreed limits within which a piece of work can proceed without escalation — typically defined across six areas in PRINCE2: time, cost, scope, quality, risk and benefit. As long as the forecast stays inside the tolerance, the manager running the work just gets on with it. An exception is raised only when a forecast shows a limit is about to be breached.

Is manage by exception the same as micromanagement?

It is the opposite. Micromanagement means a senior person checks every decision; manage by exception means they agree the boundaries up front and then leave the work alone unless those boundaries are forecast to be crossed. Done well, it gives the people doing the work more autonomy, not less, while keeping the escalation honest.

Where does the term come from?

The phrase predates PRINCE2 in general management literature, but its best-known modern form is one of the seven principles of PRINCE2, the project-management method governed by Axelos (now part of PeopleCert). In PRINCE2 it is paired with the idea of delegated tolerances and exception reports.

When should you raise an exception?

Raise an exception as soon as a forecast — not just the current position — shows a tolerance will be breached. The point is to act while options still exist, rather than reporting the breach after it has happened. A late exception is just a surprise with paperwork.

Term

RAG Status

A RAG status reports the health of a project or task with a traffic-light colour: Red, Amber or Green. It is a fast way to communicate at a glance, but on its own it is a summary of someone's judgement, not evidence of control — a green light only means as much as the owner, dated next step and fallback sitting behind it.

Term

RAID Log

A RAID log is a simple project-management tool that tracks four things in one place: Risks (what might go wrong), Assumptions (what you're taking as true), Issues (what has already gone wrong), and Dependencies (what you're relying on others for). It keeps the sources of drift visible and owned.

Guide

What Does It Mean for Work to Be 'Under Control'?

Work is under control when uncertainty has been converted into a usable control point: a clear owner, a dated next step, a decision or escalation date, and a fallback. Control is about predictability, not busyness — a controlled piece of work can still go wrong, but it can no longer surprise you.

← Back to the glossary