What is a cost-of-a-mistake ladder?
A cost-of-a-mistake ladder, a delegation method from Luba Inz of Efficiency Sprint, is a set of dollar limits, one for each level of a company. Each limit says how big a mistake that level may make on its own. Below the limit, people decide. Above it, the decision goes up a level.
Last updated:
An example ladder
Four levels, as an example. Luba ran a ladder like this as a COO; the amounts here are illustrations. Set your own.
| Level | Decides alone up to |
|---|---|
| Level 1 (top) | $5,000 |
| Level 2 | $2,000 |
| Level 3 | $750 |
| Level 4 | $150 |
How to set it
- List the levels. Start at the top and work down, following how the company really runs.
- Pick a number for each. The largest mistake you can live with at that level. It will feel like a guess. Set it anyway and adjust it later.
- Say the numbers out loud. A limit nobody has heard changes nothing.
- Turn time into money. Multiply hours lost by the hourly rate to get a dollar cost. Fifteen hours at $30 an hour is $450, which fits under a $750 limit.
What it changes
Escalations come only when a decision crosses the line. Everything under it gets settled where it starts. The owner stops getting the small calls and keeps the large ones.
From Luba Inz's work with ~100 business owners: as a company grows past 50 people, fewer decisions reach the owner, but each one is bigger. Delegation moves down a level.
A ladder is also the first fix for learned helplessness in business: people can see which mistakes are theirs to make.
Pair it with a weekly check
Without tracking, limits drift. Each week, every level reports the same short set: what got done, what worked, what didn't, and the plan for next week. Lower levels report more often. The weekly operations meeting is where the exceptions land.
Review the numbers after 30 days. If the same escalation keeps coming back, the limit is probably too low.
From ladder to matrix to AI
The ladder sets one number per level. A decision authority matrix sets it per decision: who decides, up to what amount, and what brings it back to the owner. For field services, see who approves refunds, discounts and overtime.
An AI agent can work under the same limits once they're written down. Below the line, the agent acts. Above it, the agent stops and a named person decides. You always need to know who is responsible. And it can't be AI.
In SCOPE™ by Efficiency Sprint, the ladder belongs to Ownership: people run it reliably. Pulse and Engine, where AI watches and then acts, come after.
Efficiency Sprint makes companies operationally ready for AI. The operational AI readiness check takes about two minutes and shows where your company stands.
Questions
How do I set spending limits for employees?
Start at the top level with the amount you'd approve without reading the details. Step it down level by level. Say the numbers to the team, then check after 30 days which escalations keep coming back.
What should the limit be for a team lead?
High enough that routine calls in their area never reach you. In the example ladder, the level under the top decides up to $2,000. Your number depends on your margins and volume.
What happens above the limit?
The decision goes up one level. Each level sees only what's bigger than the limit below it, so the owner sees the few decisions that are truly large.
Can an AI agent use the same limits?
Yes, once they're written down. Under the limit, the agent acts. Over it, the agent stops and the named person decides.