playbooks

home / skills / gracebotly / flowetic-app / retell

retell skill

/workspace/skills/retell

This skill analyzes conversational AI voice agent metrics and call analytics to help optimize performance and improve Retell insights.

npx playbooks add skill gracebotly/flowetic-app --skill retell

Review the files below or copy the command above to add this skill to your agents.

Files (1)
SKILL.md
210 B
---
name: retell
version: 1.0.0
platformType: retell
description: Conversational AI voice agent analytics and call metrics for Retell.
lastUpdated: 2025-12-30
---

# Retell Skill
## Templates
- voice-analytics

Overview

This skill provides conversational AI voice agent analytics and call metrics tailored for Retell. It surfaces measurable call insights, agent performance indicators, and conversation-level analytics to help teams improve voice interactions. The focus is on fast, practical metrics that drive coaching and product decisions.

How this skill works

The skill ingests voice call data and extracts key metrics such as call duration, silence periods, speech rates, sentiment, and intent triggers. It aggregates per-agent and per-skill dashboards, highlights anomalous conversations, and produces time-series trends for performance tracking. Outputs are accessible via structured reports and API endpoints for integration with analytics workflows.

When to use it

  • Monitor live and historical voice agent performance
  • Identify coaching opportunities from real call data
  • Track feature or campaign impact on conversational behavior
  • Detect unusual call patterns or regressions
  • Integrate call metrics into BI dashboards and reporting pipelines

Best practices

  • Collect consistent metadata (agent ID, skill, timestamp) with every call for reliable aggregation
  • Use rolling windows for trend analysis to smooth daily volatility
  • Combine quantitative metrics (durations, talk-time) with qualitative signals (sentiment, intents) for root-cause analysis
  • Flag and sample anomalous calls for manual review rather than trying to auto-label every edge case
  • Instrument post-call tagging to capture disposition and outcome for better metric correlation

Example use cases

  • Daily agent health dashboard showing average handle time, talk/listen ratio, and sentiment
  • Coach review queue that surfaces calls with high silence or low agent talk time
  • A/B test voice model changes by comparing pre/post trends in intent recognition and call resolution
  • Alerting on sudden spikes in call escalations or drop-offs for rapid incident response
  • Exported metrics to BI tools for executive reporting on contact center efficiency

FAQ

What types of call metrics are provided?

The skill provides duration, silence periods, talk/listen ratios, speech rate, sentiment scores, intent hits, and resolution indicators.

Can outputs be integrated with external dashboards?

Yes. Metrics are exposed via structured reports and APIs designed for easy export to BI and monitoring tools.