---
sourceDocument: Australia Enable AI
sourceDocumentLink: https://www.servicenow.com/docs/r/australia/intelligent-experiences

 Release :

    - australia

ft:locale :

    - en-US

ft:publication_title :

    - Australia Enable AI

ft:clusterId :

    - platai

bundleId :

    - platai

workflow :

    - Platform


---

# Conversation evaluation

# ServiceNow AI Conversation evaluation {#ariaid-title1}

Release version: Australia  
Updated August 3, 2026  
![](https://www.servicenow.com/docs/portal-asset/ico-clock) 3 minutes to read
Summarize  
![AI sparkle icon](https://servicenow.com/docs/portal-asset/ai-sparkle-icon) Summarized using AI  
This content was generated using new OpenAI-powered functionality. Results are provided on an as is basis and are not guaranteed to be accurate or complete.  

## Summary of ServiceNow AI Conversation evaluation

The ServiceNow AI Conversation evaluation dashboard offers comprehensive insights into the quality and effectiveness of AI-generated conversational responses.
It displays key performance indicators (KPIs), trend analyses, volume metrics, and detailed evaluation records that combine both human feedback and automated scoring.
This enables ServiceNow customers to monitor AI response accuracy, user satisfaction, and conversation quality over time.
Show full answer Show less  
Accessible via AI Control Tower under Insights, the dashboard lets users filter and analyze AI performance across multiple quality dimensions such as intent accuracy, request completion, truthfulness, and context retention. This detailed visibility supports informed decision-making for optimizing AI interactions.

## Key Features

* **Multi-dimensional Quality Metrics:** Evaluate AI responses on intent accuracy, request completion, slot filling, truthfulness (hallucination prevention), context retention, conciseness, smoothness (deadlock avoidance), and coherence.
* **Customizable Time Range:** View metrics defaulted to the last 30 days or adjust to periods like the last 7 or 90 days, or a custom date range to track AI performance trends.
* **Average Auto-evaluation Score:** Automated algorithms assess AI responses based on predefined quality criteria such as relevance, accuracy, completeness, and tone.
* **Average Human Feedback Score:** Captures user or evaluator satisfaction ratings, often reflecting qualitative factors like tone and helpfulness that automated scores may miss.
* **Evaluation Score Trend Visualization:** Line charts show weekly average satisfaction scores to help track improvements, stability, or declines in AI performance over time.
* **Total Evaluations Volume:** Bar charts display the count of AI response evaluations per week, ensuring statistical reliability and highlighting usage patterns.
* **Detailed Evaluations Table:** Provides individual records including state, human and auto evaluation scores, and score gaps, with options to toggle views for human scores and auto-evaluated labels.

## Practical Benefits for ServiceNow Customers

* Gain actionable insights into the strengths and weaknesses of AI conversational agents to improve user experience and AI accuracy.
* Identify discrepancies between human and automated evaluations to refine evaluation criteria and better align AI behavior with user expectations.
* Monitor trends and volumes to detect shifts in AI usage or performance, facilitating timely interventions or enhancements.
* Leverage detailed feedback and scoring to support continuous AI training and optimization efforts within the ServiceNow platform.  
The ServiceNow AI Conversation evaluation dashboard displays AI response quality metrics and satisfaction scores for individual conversations. It includes detailed evaluation records with human feedback and
automated scoring across quality dimensions such as intent accuracy, request completion, truthfulness, and context retention.

## Conversation evaluation overview {#md-servicenow-ai-conversation-evaluation__evaluation-overview}

The ServiceNow AI Conversation evaluation dashboard displays key performance indicators (KPIs), trend analysis, volume metrics, and detailed evaluation records. These metrics collectively provide
visibility into AI response quality and user satisfaction levels.  
To view the ServiceNow AI Conversation evaluation dashboard on AI Control Tower, navigate to AllAI Control TowerHomeInsightsServiceNow AIEvaluationConversation evaluation.  
Note:  
You can filter the Evaluation dashboard view by:

* User Satisfaction: Displays the overall satisfaction metric aggregated across all evaluation criteria.
* Request Completion: Measures whether the AI response fully addresses the user's request or question.
* Intent Accuracy: Assesses whether the AI correctly understood the user's underlying intent or purpose.
* Slot Filling: Evaluates the AI's ability to extract and identify key information entities from conversations.
* Truthfulness (Hallucination Prevention): Evaluates the AI's ability to extract and identify key information entities from conversations.
* Context Retention: Assesses whether the AI maintains and references relevant context from prior conversation turns.
* Conciseness: Measures whether the AI response is appropriately brief and focused without unnecessary verbosity.
* Smoothness (Deadlock avoidance): Assesses the AI's ability to navigate conversation flow without getting stuck or repeating itself.
* Coherence: Measures whether the AI response is logically consistent, well-organized, and easy to follow.
{#md-servicenow-ai-conversation-evaluation__ul_hkh_kkx_ckc}All metrics on the Value dashboard default to the last 30 days. Use the time period selector in the top-right corner to adjust the date range and view metrics for different time periods (for example, last 7 days, last 90 days, or a custom range).

## Average autoeEvaluation score {#md-servicenow-ai-conversation-evaluation__section_kc2_zfx_ckc}

The Average auto-evaluation score widget displays the mean satisfaction score calculated from automated evaluation algorithms applied to AI-generated responses. The auto-evaluation score uses predefined criteria to assess
whether responses meet quality standards, including relevance, accuracy, completeness, and appropriate tone.

## Average human feedback score {#md-servicenow-ai-conversation-evaluation__section_zny_dgx_ckc}

The Average human feedback score widget captures the mean satisfaction rating provided directly by users or evaluators who interact with AI responses. Human feedback scores are typically higher than auto-evaluation scores
when the AI system is performing well. Users may appreciate factors that automated systems don't capture, such as tone, helpfulness, and contextual understanding. Significant gaps between these metrics may indicate areas
where evaluation criteria need adjustment.

## Evaluation score trend {#md-servicenow-ai-conversation-evaluation__evaluation_score_trend}

The Evaluation Score Trend line chart visualizes the trajectory of average auto-evaluation user satisfaction scores over a selected time period.

You can view the average satisfaction score on the Y-axis and the weekly periods on X-axis along with the evaluation trend with fluctuation over time.

This metric helps you determine whether your AI system performance is improving, degrading, or remaining stable, enabling data-driven decisions about optimization efforts.

## Total evaluations {#md-servicenow-ai-conversation-evaluation__total-evaluations}

The Total Evaluations bar chart shows the volume of AI response evaluations completed per week across your selected time period.

You can view the number of evaluations on the Y-axis and the weekly periods on the X-axis. A consistent evaluation volume strengthens the statistical reliability of your performance metrics. Sudden drops in evaluation volume
may indicate reduced AI agent usage or fewer user interactions.

## Evaluations {#md-servicenow-ai-conversation-evaluation__evaluations}

The Evaluations table provides a detailed record-by-record view of all AI response evaluations within your selected date range. You can see the evaluations listed by State, Human user satisfaction score, Auto evaluation user satisfaction score, and Gap (Human - Auto evaluation user satisfaction score).  
Note:  
For a single evaluation record, you can select to view the evaluations by View human scores and Label auto evaluated scores, where as Label random scores is the default view.

*[\>]: and then