"What gets measured gets managed." But in voice AI, what should you actually measure? With dozens of available metrics, it's easy to drown in data without gaining insight. This guide explains every metric in our dashboard and which ones actually matter for your goals.
Outcome Metrics: The Big Picture
Containment Rate
What it is: Percentage of calls resolved by AI without human intervention.
Why it matters: Primary measure of AI effectiveness. Directly impacts cost savings.
Benchmark: 50-70% is good. 70%+ is excellent.
Customer Satisfaction (CSAT)
What it is: Post-call survey rating, typically 1-5 scale.
Why it matters: Ultimate measure of customer experience quality.
Benchmark: 4.0+ is good. 4.5+ is excellent.
First Call Resolution (FCR)
What it is: Percentage of issues resolved without customer calling back.
Why it matters: Repeat calls are expensive and frustrate customers.
Benchmark: 70%+ is good. 85%+ is excellent.
Driver Metrics: What Causes Outcomes
Intent Recognition Accuracy
What it is: How often AI correctly identifies why the customer is calling.
Impact: Low accuracy → misrouted calls → low containment, low CSAT.
Sentiment Trend
What it is: How customer emotion changes during the call (improving/declining/stable).
Impact: Declining sentiment during calls predicts low CSAT, even if resolved.
Escalation Rate by Intent
What it is: Which call types require human handoff most often.
Impact: Identifies where AI training is weakest. Priority for improvement.
Diagnostic Metrics: Troubleshooting
Average Response Latency
What it is: Time between customer finishing speech and AI responding.
Red flag: Latency >400ms feels unnatural and damages satisfaction.
Speech Recognition Confidence
What it is: How confident the AI is in what it heard.
Red flag: Low confidence clusters may indicate accent issues or audio problems.
Fallback Rate
What it is: How often AI says "I didn't understand" or uses generic responses.
Red flag: High fallback rate indicates missing training or unclear customer speech.
Building Your Dashboard
Recommended Dashboard Setup:
- Executive view: Containment, CSAT, cost savings (weekly)
- Operations view: Volume, handle time, escalation rate (daily)
- Training view: Intent accuracy, fallbacks, new intents detected (daily)
- Technical view: Latency, errors, system health (real-time)
"We were drowning in metrics. Then we focused on just three: containment, CSAT, and sentiment trend. Everything else became diagnostic only looked at when those three indicated a problem."
Operations Director Insurance Company
Need Help Setting Up Analytics?
Our customer success team can help you configure dashboards that match your goals.
Schedule Analytics Review →
