Every company sets a level above which spending needs a signature from higher up. The threshold controls cost, and it also changes the shape of what people buy.
The limit becomes a design constraint
Once a manager knows the ceiling, proposals arrive sized to fit under it. A project that would naturally cost slightly more is scoped down rather than escalated.
This is not usually deception. Escalation costs weeks, and a manager weighing a smaller purchase now against a larger one next quarter reasonably picks the one they can start.
The result is a portfolio of spending that clusters just below each threshold, which is visible in any accounts payable extract and rarely reflects the best available scope.
Splitting is the predictable response
A purchase above the limit can often be divided into pieces that are each below it. Two phases, two vendors, or a pilot followed by an extension all achieve this.
Finance teams watch for the pattern because it defeats the control while satisfying its letter. Detection usually relies on vendor totals over time rather than on individual invoices.
The behavior tends to indicate that the threshold is set too low for the work rather than that the manager is evasive. Persistent splitting is a signal about the policy.
Approval time is a hidden cost
Each escalation adds elapsed time, and the delay is rarely measured against the money the control protects. A two-week wait on a modest expense can cost more than the expense.
Senior approvers also lack context on small purchases and mostly approve them. The signature transfers accountability without adding judgment, which is the least useful form of control.
Companies that review this often find the same conclusion: raise the limit for routine categories and tighten scrutiny on the few categories where real risk lives.
Recurring commitments escape the test
Thresholds are usually applied to a single transaction, so a monthly subscription slips under the limit while its annual total would not. Software spending accumulates this way in most companies.
Renewals are worse, because they require no action. A commitment approved once continues indefinitely unless someone actively cancels it, and nobody owns that review.
Controls that evaluate annualized value rather than invoice value catch this, but they are harder to administer and less common than the simple per-transaction rule.
The threshold signals trust
Employees read their spending limit as a statement about how much judgment the company thinks they have. A limit well below their responsibility is felt as a contradiction.
Companies that widen limits while requiring visible after-the-fact reporting often get better outcomes than those that require pre-approval. Transparency substitutes for permission.
Where the limit sits therefore has a management effect independent of its financial one, and it is usually set by finance without that effect being discussed.