> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://poly-ai.ferndocs.com/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://poly-ai.ferndocs.com/_mcp/server.

# PolyScore

> Automated conversation quality scoring for voice calls.

PolyScore is an AI-powered quality score assigned to every eligible **voice call**. It evaluates the transcript and produces a normalized **0–10 rating** along with per-dimension breakdowns, helping you identify strong and weak conversations without manual review.

![polyscore](/_fern-files/_img/4b06ef4a7244753d1bd650e5c10c84096639fa82a50e818698428592075f16ad.webp)

PolyScore and Call Summaries use **GPT-5** for quality assessment, providing detailed explanations for each score dimension.

## How scoring works

PolyScore evaluates every eligible voice conversation automatically. The model reads the full transcript and rates six behavioral signals, which are combined into two sub-scores and then normalized to a 0–10 scale.

### Overall score

The overall PolyScore is a number from **0 to 10**, displayed as a color-coded badge in [Conversation review](/agent-studio/conversations/review):

| Range | Label  | Color |
| ----- | ------ | ----- |
| 7–10  | High   | Green |
| 4–6   | Medium | Amber |
| 0–3   | Low    | Red   |

### Dimensions

PolyScore evaluates three dimensions, each contributing to the overall score:

| Dimension                | Weight | What it measures                                                                 |
| ------------------------ | ------ | -------------------------------------------------------------------------------- |
| **Conversation quality** | 40%    | Whether the agent understood the user and maintained a natural conversation flow |
| **Task success**         | 40%    | Whether the user's request was resolved and the task completed                   |
| **Customer experience**  | 20%    | Whether the user had to repeat themselves or showed signs of frustration         |

Each dimension is rated as **Poor**, **Fair**, or **Good** with a text explanation. You can view per-dimension ratings and explanations in [Conversation review](/agent-studio/conversations/review).

In Agent Studio, conversation quality and customer experience are combined into a single **Agent Quality** sub-score in the UI, alongside a separate **Task Success** sub-score. The underlying three-dimension model is the same.

## Where PolyScore appears

* **Conversation review** – Score badge at the top of each transcript, with expandable dimension breakdowns
* **Conversations table** – Sortable PolyScore column for quick quality scanning
* **Home page** – Average PolyScore trend chart under Quick Insights
* **Smart Analyst** – Use PolyScore as a sampling criterion or query PolyScore tables directly via SQL
* **Conversations API** – PolyScore data is available in the API response when the conversation has been scored

## Eligibility

Not all conversations receive a PolyScore:

| Requirement       | Detail                                                                             |
| ----------------- | ---------------------------------------------------------------------------------- |
| **Channel**       | Voice conversations are scored. Webchat and SMS conversations are not scored.      |
| **Minimum turns** | At least 3 turns. Short interactions (for example, immediate hangups) are skipped. |
| **Engagement**    | If the user does not engage at all, the score is marked as N/A.                    |

## Limitations

> **Warning**
>
> PolyScore evaluates conversations based on the transcript alone. It does not have access to your knowledge base, flows, external systems, or expected outcomes.

This means:

* PolyScore **cannot verify whether an action was actually completed** in an external system (e.g., a booking made, an appointment canceled). It can only assess whether the conversation *appeared* to resolve the task based on what was said.
* PolyScore does not know what the agent *should* have said – only what it *did* say. If the agent confidently gave an incorrect answer, PolyScore may still rate the conversation highly.
* Scores reflect conversational quality, not business accuracy. Use PolyScore alongside your own QA processes and [custom metrics](/agent-studio/analytics/kp-is) for a complete picture.

## Interpreting scores

Use PolyScore as a **screening tool**, not a definitive quality judgment:

* **High scores (7–10)** – The conversation flowed well, the user's request appeared resolved, and they did not show frustration. Worth spot-checking to confirm the agent followed the correct process.
* **Medium scores (4–6)** – Some issues were detected. Review the dimension breakdowns to understand whether the problem was conversational flow, task completion, or user experience.
* **Low scores (0–3)** – Significant issues detected. Prioritize these for manual review to identify knowledge gaps, flow problems, or agent behavior issues.

Filter conversations by PolyScore in the Conversations table to build a daily QA workflow – review a sample of low-scoring voice calls to catch recurring issues early.

## Related pages

#### [Conversation review](/agent-studio/conversations/review)

View per-dimension PolyScore breakdowns alongside transcripts.

#### [Smart Analyst](/agent-studio/studio-assistant/smart-analyst)

Query PolyScore data and sample conversations by score.

#### [Studio transcripts](/agent-studio/account/call-data/studio-transcripts)

Access transcripts and call summaries.