> ## Documentation Index
> Fetch the complete documentation index at: https://docs.threetone.in/llms.txt
> Use this file to discover all available pages before exploring further.

# Success evaluation

> Author precise outcome criteria and interpret saved evaluation results without treating configuration as execution.

A success criterion states what the transcript must demonstrate. For example, a booking assistant should confirm the requested date before requesting a reservation.

<Note>
  Criteria authoring and stored-result reads are available. [Run conversation evaluation](/api-reference/agents-workspace-analytics/run-conversation-evaluation) currently returns the existing record; it does not calculate a fresh score.
</Note>

With agent edit access, review `platform_settings.evaluation.criteria`. Each criterion has a stable `id`, a `name`, and a `conversation_goal_prompt` of up to 2,000 characters.

| Field | Saved meaning |
| - | - |
| `scoring_mode` | `binary` by default, or `numeric_uniform` |
| `scope` | `conversation`, or the defining `agent`'s portion |
| `max_score` | Numeric scale ceiling, default 100 |
| `score_instructions` | Optional numeric scoring rubric |
| `llm` | Optional per-criterion model preference |

<Steps>
  <Step title="Write an observable criterion" id="evaluation-write-criterion">
    Describe the required evidence and how to handle missing information. Keep unrelated outcomes in separate criteria so one success cannot hide another failure.
  </Step>

  <Step title="Preserve the criteria list" id="evaluation-preserve-list">
    Save through the Analysis editor or [agent draft](/api-reference/agents-platform/create-agent-draft). Review the complete criteria list and identifiers before saving a replacement.
  </Step>

  <Step title="Review stored results" id="evaluation-review-results">
    Read `analysis.evaluation_criteria_results` in [conversation details](/api-reference/agents-platform/get-conversation-details). Inspect each result and rationale, plus `score` and `max_score` when present. Compare them with the relevant transcript turns.
  </Step>
</Steps>

<AccordionGroup>
  <Accordion title="Is an unknown result a failure?" id="evaluation-faq-unknown">Keep unknown separate from success and failure. It does not provide enough evidence to claim the criterion passed.</Accordion>
  <Accordion title="Does changing a criterion rescore history?" id="evaluation-faq-history">No. Saving a definition does not recalculate existing conversations.</Accordion>
</AccordionGroup>


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.