Twilio - Insights
Twilio - Insights
Twilio Voice Insights surfaces call-quality analytics — packet loss, jitter, MOS scores, per-participant diagnostics, and issue attribution for debugging WebRTC/SIP call quality.
Full RTP-layer metrics (packet loss, jitter, MOS)
Voice-only — does not cover SMS/messaging
GET /Voice/Calls/{CallSid}/Events for the per-call event stream, GET /Voice/Calls/{CallSid}/Metrics for RTP metrics. Console also has Insights Dashboard.
Uptime · 30-day window
About this API
WebRTC/SIP call quality issues are notoriously hard to reproduce and attribute — is it the user's home WiFi? The link between Twilio and the carrier? An SDK bug? Voice Insights surfaces fine-grained RTP-layer metrics and event tags including MOS (Mean Opinion Score for subjective quality), packet loss, jitter, and specific Issue labels (high_jitter, constant_audio_level). Each call has independent data streams for both parties, so you can tell whether the issue is one-sided or two-sided. For teams running Twilio Voice in production, Insights is a required observability layer. Pricing is per-call (Insights billed separately when enabled).
What you can build
- 1Investigate customer complaints of "I couldn't hear them"
- 2Monitor WebRTC client-side call quality
- 3Pinpoint root cause (client network vs Twilio platform)
- 4Generate quarterly call-quality reports
Strengths & limitations
Strengths
- Full RTP-layer metrics (packet loss, jitter, MOS)
- Root-cause attribution across client / carrier / Twilio platform
- 30-day historical retention for queries
Limitations
- Voice-only — does not cover SMS/messaging
- Beyond 30 days requires self-managed archival
Example request
curl https://support.twilio.com/<endpoint>Getting started
GET /Voice/Calls/{CallSid}/Events for the per-call event stream, GET /Voice/Calls/{CallSid}/Metrics for RTP metrics. Console also has Insights Dashboard.
FAQ
Is Insights on by default?+
No. Enable it per-call (Voice Insights parameter) or set account-default. Each call incurs separate billing.
What MOS counts as "good"?+
4.0+ is good, 3.5-4.0 acceptable, below 3.0 noticeably bad. Twilio targets >= 4.0 on its side.
Can I access raw RTP packet data?+
No. Insights exposes second-level aggregated metrics, not packet captures.
Technical details
- Auth type
- unknown
- Pricing
- unknown
- Protocols
- REST
- SDKs
- python, javascript, java, ruby, php, csharp, go
- Response time
- 144 ms
- Last health check
- 5/12/2026, 7:38:32 AM
More from Twilio
Manage Twilio account credentials and hierarchy — create subaccounts, rotate auth tokens, and look up account metadata.
Twilio's core API for voice calls, SMS, Verify, and other programmable communications, billed per usage.
Twilio Autopilot is a sunset conversational-AI platform formerly used for IVR bots and NLU tasks (now superseded; migrate to Voice Intelligence or third-party LLMs).
Bulk-export Twilio usage history (Messages, Calls) as compressed daily JSON files — for audit, billing reconciliation, or data-warehouse ingestion.
Twilio Chat v1 is a sunset real-time chat SDK backend (superseded by Conversations). Existing projects still work; new projects should use Conversations API.
Twilio Content API centrally manages messaging templates across channels (WhatsApp/SMS/Messenger), including WhatsApp template approval flow and dynamic variable binding.
Twilio Conversations unifies multi-channel (SMS/MMS/WhatsApp/Chat/Email) messaging into a single "conversation" primitive with multi-participant, persistence, and read receipts.
Twilio Events streams Twilio platform events (message delivery, call state changes, quota alerts) via a unified webhook bus — replacing scattered per-endpoint webhook configs.