--- title: "Call Logs & Analytics - Auto-Respond Docs" description: "Review every call your agents have made - transcript, recording, AI summary, extracted data, and appointment details." source: https://auto-respond.com/docs/voice-agent/call-logs-analytics/ last_updated: 2026-04-21 --- ![Call Logs & Analytics documentation guide](https://auto-respond.com/assets/docs/voice-agent-call-logs-analytics.svg "Call Logs & Analytics Documentation") A visual guide to the Call Logs & Analytics documentation, covering the setup, configuration, and workflow details explained on this page. Documentation illustration for Call Logs & Analytics. # Call Logs & Analytics Every call your agents place or receive is recorded in the Call Logs. Each log entry has the full transcript, audio recording, an AI-generated summary, the customer data extracted from the call, the appointment booked (if any), disconnection reason, and duration. ## Getting here In Auto-Respond, go to **Voice Agent** → **Call Logs**. ## List view The list shows every call, newest first. Pagination is 50 calls per page. ### Filters - **Agent** - filter to calls handled by one specific agent. - **Date range** - preset options or a custom range. - **Status** - one of: `initiated`, `ringing`, `answered`, `completed`, `failed`, `busy`, `no-answer`, `cancelled`. - **Direction** - `inbound` or `outbound`. - **Batch call** - filter to calls from a specific batch campaign. - **Sentiment** - positive / neutral / negative (when sentiment extraction is available for the call). - **Unread only** - hide calls you’ve already reviewed. ### Mark read / unread - Every new call is unread until you open it. - Click the read/unread indicator on any row to toggle it manually. - **Mark all as read** respects the current filter. Filtering to a single agent and clicking it marks only that agent’s calls as read. ### Sort Default sort is newest first. This is not currently configurable. ## Detail view Click any call to open the detail view. ### Transcript The full conversation, line by line, with speaker turns labeled (agent vs. customer). Timestamps are included for each turn so you can jump to a specific moment in the recording. ### Recording Audio playback of the full call. Play, pause, scrub, or download. ### AI-generated summary A short write-up produced after the call finishes analysis. Typically one to three sentences covering what happened and the key outcome. Includes: - **Callback: Yes / No** - whether the customer’s need was resolved on the call, or whether a human follow-up is required. - A brief narrative summary (e.g., _“Customer needed a water heater repair. Booked a technician for Thursday afternoon. Customer preferred text confirmation.”_). ### Extracted data Structured data pulled out of the conversation by the AI: - `customer_name` - `callback_phone` - `customer_email` - `customer_address` Fields that didn’t come up on the call are left empty. ### Appointment booked If the agent booked an appointment during the call, the booking details are shown here - date, time, service, and any customer notes attached to the booking. ### Disconnection reason The reason the call ended - e.g., `customer_hangup`, `agent_hangup`, `transfer_completed`, `no_answer`, `voicemail_reached`. Useful for diagnosing calls that ended unexpectedly. ### Duration Total call length in seconds or minutes. ## Email notifications Each agent has a **Send email on call analyzed** toggle in the Advanced tab of its configuration. - When on: after a call is analyzed (a few seconds after it ends), a summary email is sent to the email address you configure. The email includes the AI summary, extracted data, and a link back to the call in the log. - When off: no email is sent. You’ll still find the call in the list. See [Agent Configuration](https://auto-respond.com/docs/voice-agent/agent-configuration/). 💡 Turn this on for Calling Agents and Batch Calling Agents so you get near-real-time visibility into outbound results. For Answering Agents, it’s optional - opening the Call Logs once a day is often enough. ## Common workflows ### Daily review of inbound calls 1. Open **Call Logs**. 2. Filter by direction `inbound` and the past 24 hours. 3. Filter **Unread only**. 4. Skim the AI summary on each. Open the transcript only for calls flagged “Callback: Yes” or with unusual sentiment. 5. Click **Mark all as read**. ### Finding missed opportunities 1. Filter by **Status** `no-answer`. 2. Sort through the list and follow up manually on high-value leads. 3. For recurring no-answers, check your **Timing** (call hours) and **Voicemail Message** - a better voicemail lifts callback rates. ### Reviewing a batch campaign 1. Open **Call Logs**. 2. Filter by the **Batch call** name. 3. Review the Callback: Yes flags to see which customers agreed to a booking or asked for a callback. 4. Export customer data for anyone flagged for human follow-up. ## Troubleshooting **A call I just made isn’t in the list.** Analysis takes a few seconds to a minute after the call ends. Refresh the page. **The transcript is empty.** Very short calls (under a few seconds) may not produce a usable transcript. Check the recording instead. **The summary says “Callback: Yes” but the customer booked during the call.** Open the transcript - the agent may have promised a follow-up (for a part check, address confirmation, etc.) that’s separate from the booking. The flag means the customer’s need wasn’t fully resolved. **Sentiment is missing.** Sentiment isn’t extracted for every call. It’s shown when available and filterable when present. **The “Extracted data” fields are blank.** The agent only fills fields that the customer actually provided. If the caller didn’t say their address, `customer_address` is empty - that’s expected. *** Questions? Email . Last updated: April 21, 2026