Approval Limits and Controls: Thresholds, Who Approves What, and How It Is Recorded

In many small businesses everything goes to one person for approval - so everything waits for them. The fix is not abandoning control but moving it from a case-by-case decision into a rule: what can be approved without asking, up to what amount, under what conditions.

What actually needs approval

Row four is often forgotten and is among the most significant: agreeing long payment terms is a cash-flow decision made in a sales conversation, sometimes without anyone noticing.

How to set a threshold

Not by instinct. Three questions decide it: what amount is one where an error is still tolerable; how often this happens per month; and how much time the approval itself costs. When the approval costs more than the risk, the threshold is too low.

In practice, most businesses find that most approvals concern small amounts, and that the bulk of the time goes precisely on those. Raising the threshold slightly - so most cases need no approval - returns most of the time and leaves control exactly where it is needed. The logic of a human bottleneck is covered in finding the bottleneck in your process.

What must stay with the owner

• A new supplier - because it is an ongoing commitment rather than a one-off spend.
• An unusual price for a regular customer - because it becomes a precedent.
• A legal commitment - a contract, a guarantee, unusual terms.
• Hiring and salary commitments.
• Anything above the high threshold you defined.

What has no reason to stay: approving office supplies, approving a small credit, approving a discount of a few per cent. Those are precisely the cases where approval costs more than the amount.

Recording - what is enough

No system is needed. What is needed is three things per exceptional approval: who approved, when, and on what exactly. A message in the work channel is sufficient, provided it is linked to the order or customer rather than sitting in a private conversation.

What matters is that the record is somewhere findable in six months. An approval given verbally and never recorded is a source of argument: the rep remembers it was approved, the manager remembers hesitating, and there is no way to settle it - which is exactly the problem described in task handoffs without drops.

How to tell the system is working

Three signals. How many approvals a week - if the number did not fall after you set thresholds, the thresholds are too low. How long people wait for approval - the metric the team feels. And how many approvals were refused - if the answer is zero over two months, that approval is ceremonial and can be removed.

The third matters most and almost nobody measures it. An approval granted 100% of the time prevents nothing; it only adds waiting, and sometimes creates a false sense that somebody genuinely checked.

What happens when the approver is unavailable

This is the failure that makes good thresholds worthless. With no defined deputy, one person's day off stops orders, credits and quotations.

The fix is one rule: every authority has a named deputy, and the substitution triggers automatically after a defined time - if no approval within X hours, the deputy may approve. It sounds risky and is usually safer than the current state, in which people improvise anyway when the approver is away.

How do you define this in an hour?

No policy document required. One page with four columns is enough: type of decision, who may decide, up to what limit, and what must be recorded. Five to eight rows cover most day-to-day decisions in a small business.

The fast way to build it is not to plan ahead but to look back: review the last fortnight and list everything that actually required approval. That list is almost always short, and usually three kinds of decision explain most of the requests - and thresholds for all three can be written in ten minutes.

Once the page exists, put it somewhere visible rather than emailing it. An authority you have to search for in order to know it exists will be ignored, and people will go back to asking - which is exactly what you were trying to prevent.

What happens when you grant authority and nobody uses it

This happens more often than expected, and it is a habit problem rather than a trust one. A team used to asking keeps asking, even when entitled to decide - mainly because prior approval protects them if something goes wrong.

What changes that is not the permission but the reaction: when somebody decides within the threshold and the outcome is imperfect, the reaction determines whether they will decide again. A reaction that asks whether the decision was reasonable given what was known - rather than whether the outcome was good - is what produces a team that decides.

It also helps to say it explicitly, once: "if you decided within the threshold, that was the right decision even if the outcome was poor". One sentence like that, said before the first such case rather than after it, saves months of people coming back to ask about everything.

Sources

• How to improve business efficiency
• Israel Tax Authority - information for businesses