Docs navigation

Audit Trail vs Outcome Ledger: What Ran Is Not What Worked

An audit trail records what an agent did. An outcome ledger records whether it was right. Most agent tooling ships the first and calls it accountability — this is why the second is the harder, more valuable record.

Published August 16, 2026

Two records look almost identical and are not the same thing. An audit trail tells you what an agent did. An outcome ledger tells you whether it was right. The entire agent-tooling category ships the first and, with a straight face, calls it accountability. The gap between the two is where the value — and the honesty — actually lives.

What an audit trail is good for

An audit trail is a time-ordered record of actions and actors: this ran, then this, approved by that person, touching that resource. It is genuinely necessary. It answers “did anything improper happen” and “who touched this,” and a system without one cannot be trusted at all. We keep a rigorous one, append-only, so history cannot be silently rewritten. But notice the ceiling: an audit trail can record a flawless year that quietly lost money, because it never once asks whether an action worked.

What an outcome ledger adds

An outcome ledger pairs every action with the result it was supposed to produce, graded against a number chosen before the action ran. “This ran” becomes “this ran, it was betting on that, and here is whether it landed.” That second clause is the one that changes how a company operates, because it is the only clause a buyer can use to decide whether the agent was worth having. The campaign we killed and the coupon that beat its baseline are two entries from exactly this ledger.

Why the harder record is the valuable one

Building an outcome ledger is uncomfortable on purpose. It requires measurement at birth, an append-only substrate so a verdict cannot be revised into a win after the fact, and the plain willingness to count misses. Systems avoid it because it makes them provably wrong sometimes. That is precisely the property that makes it trustworthy: a record that can be wrong, and shows when it was, is a record you can believe when it says it was right.

The floor and the point

So the relationship is simple. The audit trail is the floor every serious harness must stand on. The outcome ledger is the point — the reason any of the activity mattered. When you are evaluating an agent product, find out which one you are actually being sold. If all you can see is what ran, you are looking at a floor being sold as a building. The deeper mechanics of how we trace all this live in Telemetry: we trace decisions, not calls.

Questions founders ask

What is the difference between an audit trail and an outcome ledger?
An audit trail answers "what happened and who did it" — a time-ordered record of actions, actors, and approvals. An outcome ledger answers "was it the right thing" — each action paired with the result it was supposed to produce, graded against a prediction made before it ran. The audit trail proves the agent behaved; the outcome ledger proves it helped. You need both, and almost everyone ships only the first.
Isn't an audit trail enough for accountability?
For compliance, often yes — it shows nothing improper occurred. For running a business, no. An audit trail will happily record a year of well-behaved actions that collectively lost money, because it never asks whether any of them worked. Accountability to a regulator is "we followed the rules." Accountability to an owner is "we moved the number." The outcome ledger is the second kind.
Why is an outcome ledger harder to build?
Because it forces the system to commit to falsifiable claims. An audit trail only has to observe; an outcome ledger has to predict before acting and then grade itself honestly afterward — including logging the times it was wrong. That requires measurement at birth, an append-only record so verdicts cannot be quietly revised, and the discipline to count misses. It is uncomfortable by design, which is why most tools stop at the audit trail.
Does Figaro keep an audit trail too?
Yes — the outcome ledger is built on top of one. Every action still records who or what proposed it, who approved it, and what it touched, on an append-only substrate where corrections are new rows rather than edits. The difference is that we do not stop there: each action also carries the metric it promised and the verdict it earned. The audit trail is the floor; the outcome ledger is the point.
Drafted by the Figaro content seat · edited by Fable · reviewed by Kyle · last updated August 16, 2026