Recovery distance
The amount of human and system work needed to get from an AI failure back to a safe, correct state — a cost that belongs in the business case, not just the demo.
Every AI system fails sometimes. What separates a tolerable failure from an expensive one is how far you have to travel to get back to a correct state. Recovery distance measures that trip: how many people, how many steps, how much time and how much system effort it takes to detect the failure, undo its consequences and restore normal operation.
A failure that is detected immediately, contained to one record and reversed with one click has a short recovery distance. A failure that is noticed days later, has already propagated into downstream systems and needs manual reconciliation has a long one, even if the original error was small.
Why it matters
Vendor demos show the cheapest path through the system: success. A business case built only on that path ignores the cost of the unhappy path. Putting recovery distance in the model forces the real question: when this goes wrong, what does it take to make it right, and who pays?
Recovery distance is also a supervision cost, because someone's attention is spent on the repair. Ask vendors for a recovery contract so the distance is written down before purchase.
Read more in Make the AI vendor demo fail.
Related terms
Recovery contract
A short written artifact in an AI vendor evaluation that states how failures are detected, contained, reversed and handed back to people — and who is responsible at each step.
Supervision load
The human attention an AI system still consumes — context supply, review, approval, correction and exception handling — measured per dependable, completed unit of work.
Dependable outcome
A result from an AI-assisted workflow that is actually usable without further repair — the right denominator for measuring the human attention and cost AI automation really consumes.