# Agent Analytics metrics and optimization

*Agentforce Specialist Study Guide · Governance and Observability · Last updated October 15, 2026 · https://www.certifyforce.com/study-guide/agentforce-specialist/governance-and-observability/agent-analytics-metrics-and-optimization*

What each Agent Analytics metric measures (deflection, escalation, abandonment, engagement, success, task resolution), how LLM-judged quality scores point to retrieval or response problems, how intents are extracted and clustered, and the loop that turns a weak subagent or intent cluster into a measured fix.

- Certification: Salesforce Agentforce Specialist
- Exam section: Governance and Observability
- Study guide: https://www.certifyforce.com/study-guide/agentforce-specialist

**Official exam objective:**

> Explain agent analytics and agent optimization.

## What the exam expects

- Define the effectiveness metrics (deflection, escalation, abandonment, engagement, success, task resolution) and read them together.
- Interpret the quality score and its dimensions (answer faithfulness, answer relevance, context relevance) and know they are judged by an LLM.
- Explain intents, the intent and clustering pipelines and their timing.
- Use Performance Insights and quality score reasoning to find the subagent or intent behind poor results.
- Choose the optimization action (knowledge gaps, overlapping articles, unclear instructions) and re-measure.

Key terms: **deflection rate** is the share of sessions that ended without escalation; **task resolution rate** is the share judged fully resolved by the Trust Layer Task Resolution detector; **answer faithfulness** measures how closely a response sticks to grounded source data; **context relevance** measures whether the right source was retrieved; **intent** is what the user was trying to accomplish within a set of turns.

## Key facts

| Fact | Detail | Source |
|---|---|---|
| Abandonment timeout | Default 2 hours | [Monitor Agent Metrics and Scores](https://trailhead.salesforce.com/content/learn/modules/agent-analytics-and-monitoring/monitor-agent-metrics-and-scores) |
| Quality Score scale | 1–5, from an LLM evaluator | [Monitor Agent Metrics and Scores](https://trailhead.salesforce.com/content/learn/modules/agent-analytics-and-monitoring/monitor-agent-metrics-and-scores) |
| Dimension scales | Faithfulness, answer relevance and context relevance are 0–1 | [Monitor Agent Metrics and Scores](https://trailhead.salesforce.com/content/learn/modules/agent-analytics-and-monitoring/monitor-agent-metrics-and-scores) |
| Score prerequisites | Audit and Feedback plus Knowledge/RAG Quality Data and Metrics | [Monitor Agent Metrics and Scores](https://trailhead.salesforce.com/content/learn/modules/agent-analytics-and-monitoring/monitor-agent-metrics-and-scores) |
| Intent pipeline | About 4–5 hours after a session closes | [Investigate Agent Sessions and Intents](https://trailhead.salesforce.com/content/learn/modules/agent-analytics-and-monitoring/investigate-agent-sessions-and-intents) |
| Clustering pipeline | Weekly; tag needs 10+ similar intents | [Investigate Agent Sessions and Intents](https://trailhead.salesforce.com/content/learn/modules/agent-analytics-and-monitoring/investigate-agent-sessions-and-intents) |
| Segment comparison | Performance Insights by subagent, intent or action | [Monitor Agent Metrics and Scores](https://trailhead.salesforce.com/content/learn/modules/agent-analytics-and-monitoring/monitor-agent-metrics-and-scores) |
| Data foundation | Built on the Session Tracing Data Model | [Agent Analytics](https://help.salesforce.com/s/articleView?id=ai.generative_ai_agent_analytics.htm&language=en_US&type=5) |

## Check yourself

- Why can an agent show high deflection and still leave customers unresolved?
- When will a new cluster tag appear for a brand-new kind of request?
- Which field explains why a session got a low quality rating?
- Which three dimensions make up the quality view besides the overall score?

## Sources

- [Monitor Agent Metrics and Scores](https://trailhead.salesforce.com/content/learn/modules/agent-analytics-and-monitoring/monitor-agent-metrics-and-scores)
- [Investigate Agent Sessions and Intents](https://trailhead.salesforce.com/content/learn/modules/agent-analytics-and-monitoring/investigate-agent-sessions-and-intents)
- [Get Started with Agent Monitoring](https://trailhead.salesforce.com/content/learn/modules/agent-analytics-and-monitoring/get-started-with-agent-monitoring)
- [Agent Analytics](https://help.salesforce.com/s/articleView?id=ai.generative_ai_agent_analytics.htm&language=en_US&type=5)
- [Set Up Agent Analytics](https://help.salesforce.com/s/articleView?id=ai.generative_ai_agent_analytics_setup.htm&language=en_US&type=5)

## About the full unit

The "Core concepts", "Exam traps" and "Worked scenario" sections of this unit, its 16 linked practice questions, and unlimited timed mock exams are included with the Agentforce Specialist pack (USD $19): https://www.certifyforce.com/practice-exams