Escalation rules that fire on the right things: tolerances, consequences, accumulation and authority that expires

Escalation can fail in two opposite ways. The owner can be pulled into routine decisions, slowing everything down and teaching managers that nothing is really theirs. Or the owner can stay distant until a problem has grown beyond anyone’s ability to fix it cheaply. Most businesses try to avoid both with a simple rule: managers handle anything within a certain budget or schedule tolerance, and anything bigger comes upward.

That rule is necessary, but on its own it misses three important things. It measures how far something has moved from plan, not how much it matters. It looks at one problem at a time, so several moderate problems with a shared cause never reach the owner, even when together they are serious. And it treats delegated authority as permanent: once someone is put in charge of a job, their authority runs until the job ends, with no point at which it must be renewed.

This article explains how to design escalation around tolerances, consequences and accumulation, how to give delegated authority a natural expiry, and how to decide in advance what must not be traded when objectives collide. It is general information for owners and managers who delegate work and want to hear about the right things at the right time.

Tolerances: freedom within limits

A tolerance is a limit within which the person running something can act without asking. Good tolerances:

• Cover more than money and time. Scope, quality, risk, customer impact and expected benefits can all need limits.
• Reflect the consequence of deviation. A one-week slip may be harmless on one job and commercially critical on another, so a single percentage for everything rarely works.
• Are set by the person granting authority, not proposed by the person receiving it. A limit the holder chose tends to be calibrated to what they expect to need.
• Trigger on forecast, not actual, breach. The value comes from hearing when a limit is likely to be exceeded, while there are still options.
• Require notification as a rule. If escalation depends on the holder’s judgement of how serious things are, it relies on the person who might look worst by escalating. A breached tolerance should oblige notification regardless.

Tolerances that are too tight create constant escalation that gets ignored. Too wide, and preventable losses build up before anyone hears.

Consequence triggers: when size is not the point

Variance tells you how far performance has moved. It does not tell you how much the movement matters. A two per cent delay caused by evidence of a safety problem may need the owner’s attention immediately; a ten per cent delay on a low-stakes internal task may not.

So add consequence triggers that escalate regardless of size:

• a credible risk of injury or harm;
• a possible breach of law, licence or regulation;
• a threat to a key customer relationship or the business’s reputation;
• the failure of an assumption the plan depends on;
• a benefit falling below its minimum acceptable level;
• the need to change a commitment made to someone outside the business;
• an irreversible decision with material future consequences;
• a conflict between objectives that needs a trade-off above the manager’s authority.

A simple matrix helps decide the route:

Accumulation: many moderate risks with one cause

An escalation rule that looks at each risk on its own reports upward the kind of risk least likely to sink a business: the single large exposure, which is usually visible, owned and already getting attention. It is silent about the kind more likely to: several moderate exposures driven by one or two common causes and capable of arriving together.

Businesses often justify wider tolerances at higher levels on the basis that problems in one place will be offset by good results elsewhere. That only works if exposures do not move together. In a business concentrated on a few customers, sites, suppliers, seasons or people, they often do. Moderate risks with a shared driver reinforce each other rather than cancel out.

A simple accumulation check runs monthly on the risks below the escalation line:

1. Group by cause, not by owner. Which risks trace back to the same driver: weather, a key customer, an energy contract, one supplier, one key person, one busy period?
2. Add up each group without discounting each item separately, because the point is that they would happen together.
3. Check timing. If three or more could happen in the same few weeks, escalate the group regardless of size.
4. Ask for the offset. Anyone holding a wider tolerance on the grounds that risks will balance out should be able to name what balances what, and why those things would not move together.

There is also a constraint escalation rules never measure: attention. When several moderate issues arrive at once, the scarce resource is usually not money but the number of serious decisions a small team can make well in one week.

Authority with an expiry date

Delivery authority is usually granted once: someone is put in charge of a project, and from then on they are understood to be running it until it ends. That is an unusual way to delegate. Nobody would give a spending authority without a limit or a review date. Yet the authority to commit a business’s scarcest people to months of work is often handed over without either.

An older project governance approach, described by G. J. Rankins in a comparison of two well-known project methods, offers a useful alternative. Authority is granted for a stage, a bounded period with a detailed plan, and it lapses in two ways:

• Normally, at the end of the stage, when it must be formally renewed.
• Early, if a tolerance is breached, when the manager must show the job is on track or can be brought back before more work is done.

Rankins distinguished phases, natural divisions of the work that the manager controls, from management stages, the points where authority lapses and the person who granted it decides whether to renew. A plan with phases but no stages has a schedule, not a control structure.

The appeal is that the default flips. Nobody has to take action to intervene; someone has to make a case to continue. The method also limited detailed planning to the stage in hand, often about three months ahead, on the reasoning that detailed plans further out cost effort, will be rewritten and create an impression of commitment that nobody has actually decided.

Renewal takes time, so apply it in proportion: full renewal for significant jobs above an agreed size, a lighter check below it. Applied everywhere, it degrades into ceremony.

Decide what cannot be traded

When objectives collide, someone will make a trade-off. The question is whether it is made deliberately by someone with authority or made by default by whoever is under the most pressure. Before pressure arrives, sort objectives into three classes:

• Protections that must not be traded through ordinary decisions: safety, legal compliance, ethical obligations. “Non-negotiable” does not mean impossible to change; it means changing it requires higher authority and stronger evidence.
• Minimum outcomes below which the work no longer makes sense: a service level, a customer commitment, an economic threshold.
• Variables that can be adjusted to protect the first two: timing, cost, scope detail, sequence.

A business reveals its real priorities through its trade-offs. If safety is said to come first but schedule wins every time they conflict, schedule is the real priority. The risk appetite, tolerance and capacity article covers setting these limits at the level of the whole business.

Escalate decisions, not problems

A useful escalation is a decision package, not a problem description. It states:

• what has changed;
• which limit or protection is at risk;
• what happens if nothing is done;
• the options and their trade-offs;
• a recommendation;
• what authority is needed;
• when the decision must be made.

And treat escalation as the control system working. If people are criticised for escalating, they will stop, and problems will arrive late. The decision rights before meetings article covers setting out who decides what in the first place.

A worked example

This is an illustration. A wholesale plant nursery supplies garden centres and one large hardware chain. Its manager can approve any single issue up to $30,000 without the owner. Five risks for the spring peak are each rated moderate, and none comes near that limit:

• water restrictions forcing extra carting of water: about $20,000;
• heat damage to stock: about $25,000;
• lower picking productivity and overtime in hot weather: about $12,000;
• higher power costs for irrigation pumps and cooling: about $8,000;
• the hardware chain cutting orders if stock quality falls: about $18,000.

Grouped by cause, all five trace back to one driver: a hot, dry spring. Together they come to about $83,000, and all could happen in the same six weeks. Under the old rule, the owner would have heard about none of them. Under the accumulation check, the group is escalated in winter. The owner and manager agree to install shade cloth on the most exposed beds, pre-book a water carter, adjust picking hours in heatwaves and discuss quality allowances with the hardware chain in advance.

The same review sets clearer rules for the manager’s other big job, a new greenhouse. Authority is granted for the first stage, the structure, with tolerances of two weeks and $15,000. At the end of that stage, the manager presents progress and the plan for fit-out, and the owner renews authority for the next stage. Staff heat safety is recorded as a protection that cannot be traded for output, with a clear rule for when outdoor work stops, and supplying the hardware chain is recorded as a minimum outcome.

How this applies to a small Australian business

• Set tolerances for cost, time, quality, risk and customer impact, not just budget.
• Escalate on forecast breach, and make notification a rule.
• Add consequence triggers for safety, legal, reputational and irreversible matters.
• Run an accumulation check monthly, grouping risks by cause.
• Grant authority for stages on significant jobs, with renewal at each boundary.
• Sort objectives into protections, minimum outcomes and variables before pressure arrives.
• Ask for decision packages, not problem descriptions.
• Thank people for escalating early.

Signals worth watching

• The owner approving routine items while serious issues arrive late.
• Several moderate risks sharing one cause and no one adding them up.
• Projects running for months without any formal renewal of authority.
• Escalation that depends on someone’s personal judgement of seriousness.
• Objectives that are all described as top priority.
• People reluctant to escalate because of how it was received last time.

Common mistakes

• Using one percentage tolerance for very different kinds of work.
• Escalating only after limits are breached.
• Ignoring consequence because the dollar amount is small.
• Looking at risks one at a time in a business concentrated on a few dependencies.
• Granting authority without an expiry or renewal point.
• Leaving trade-offs to whoever is under most pressure.

Frequently asked questions

Won’t this make managers escalate everything? Clear tolerances usually reduce escalation of routine matters, because managers know what they can decide. Consequence and accumulation triggers then bring up the few things that really need the owner.

How do we set the right tolerance? Start from what deviation would cost or harm, not a standard percentage. Review after a few months based on what was escalated and what should have been.

Is stage-by-stage renewal too bureaucratic for a small business? Not if applied only to significant jobs. A short meeting at the end of each stage to confirm progress and the next plan is usually enough.

What counts as a common cause? Anything that could make several risks happen together: a season, a customer, a supplier, a person, a system, a location or a price.

Who should own the accumulation check? Someone who can see across the business and act on the result, often the owner or a senior manager, with input from whoever keeps the risk list.

Questions to ask

• What can each manager decide without us, and what must they tell us about?
• Which consequences should come to us regardless of size?
• Which of our moderate risks share a cause, and what do they add up to?
• When does each manager’s authority on current major jobs next need renewal?
• What will we never trade, and what are we willing to adjust?
• Do people feel safe escalating early?

Bringing it together

Escalation rules usually fire on the size of one problem against a budget or date. That is necessary but incomplete. Add consequence triggers so serious matters reach the owner whatever their size, run an accumulation check so moderate risks with a common cause are added up, and give delegated authority stages with renewal points so it does not run unchecked. Decide in advance what must not be traded when objectives collide, and ask for escalations as decision packages. The aim is not more escalation, but escalation of the right things, early enough to act.

---

Source: KEVOS notes, drawing on teaching material on tolerances, management by exception and program governance, and on G. J. Rankins’ comparison of two project management methods and their approaches to stages and delegated authority. Examples and figures in this article are illustrations. This article is general information.