Audience
- Managers reviewing call performance
- Operators validating trends and quality outcomes
Prerequisites
- Access to analytics view
- Sufficient historical call activity for meaningful trends
What analytics typically show
- Call volume and distribution patterns
- Outcome/performance indicators from completed interactions
- Quality/coverage signals based on available data
Data latency and coverage expectations
- Some metrics may lag behind real-time operations.
- Recent calls may appear in logs before analytics aggregates update.
- Low-volume workspaces may show sparse or noisy trends.
Interpretation guidance
- Use trends over single-point snapshots.
- Compare like-for-like time windows when diagnosing changes.
- Pair analytics insights with call log spot checks for root-cause analysis.
Drill into the calls behind a metric
On the Call Insights tab, every metric is clickable — you do not have to take the number on trust. Select a property in Top Properties, an intent in Intent Signals, or a row in Top Inquiries or Top Conversational Frictions, and a window opens listing the exact calls behind that count. For each call you see:- Date — in your agent’s timezone.
- Caller — the contact’s name when we can match the number, otherwise the phone number itself.
- Summary — a short recap of what the call was about.
- View Call — opens the full call log, transcript, and recording for that call.
Long lists scroll inside their card
When a workspace has many properties, inquiry types, or friction tags, the Top Properties, Top Inquiries, and Top Conversational Frictions cards scroll within a fixed height rather than stretching down the page. Scroll inside the card to see the full ranked list — nothing is truncated.SMS metrics
The SMS tab summarises your text-message activity for the selected date range. It opens by default and shows three headline numbers plus a daily trend chart. Use the date-range control at the top to switch between presets or set a custom window — every number and the chart update together.- Sent — every text message your workspace handed to the network in the window, whether it went out from a campaign, was typed by an operator in the inbox, or was sent by your AI agent. A message is counted on the day it was actually sent, in UTC.
- Replied — every inbound text your contacts sent back in the window, counted on the day it arrived (UTC). This includes opt-out replies. When a contact texts back
STOP,UNSUBSCRIBE, or a similar keyword, that message is a reply and is counted here as well as under Unsubscribe — so a single opt-out shows in both numbers. The KPI tile carries an “Includes opt-out replies” note as a reminder. - Unsubscribe — the number of contacts who opted out of SMS in the window (for example by replying
STOP), counted on the day they opted out.
Reading the SMS numbers
- Because an opt-out is both a reply and an unsubscribe, Replied will always be at least as large as Unsubscribe for the same window. This is expected — Replied measures inbound volume, Unsubscribe measures intent.
- Days with no activity show as zero on the chart rather than gaps, so a quiet weekend reads as a dip to zero, not a break in the line.
- A brand-new or low-volume workspace may show all zeros. When every number is zero for the selected range, the tab shows a short “no activity” message instead of empty cards.
Data caveats
- Missing/partial data can affect metric confidence.
- Feature enablement (recording/transcript/config) influences available analytics depth.
- Subscription/workspace context can affect visibility of some analytics surfaces.
Troubleshooting
Analytics appears empty
- Confirm date range and workspace selection.
- Verify there were completed calls in that period.
- Wait for aggregation delay and refresh.
SMS tab shows all zeros
- Confirm the date range covers a period when you actually sent or received texts.
- Sent counts only messages that reached the network, so drafts and failed sends do not appear.
- A workspace that has never run an SMS campaign or inbox conversation will show zeros until its first message.
Numbers differ from call logs
- Confirm same time window and timezone are used.
- Check if logs are near-real-time while analytics is batched.

