Call Analytics & Quality
Understanding call quality metrics, sentiment analysis, and conversation intelligence.
Turn Every Call Into Actionable Data
Every call your AI agent handles generates a wealth of data ā and the analytics dashboard turns that raw data into insights you and your clients can actually use. Whether it's spotting trends, identifying training gaps, or proving ROI to a skeptical client, analytics is where the magic happens.
Analytics Dashboard Overview
The analytics dashboard (available in both your agency portal and the client portal) presents six core metrics at a glance:
| Metric | What It Measures | Why It Matters |
|---|---|---|
| Call Volume | Total calls in the selected period | Shows how busy the line is and trends over time |
| Quality Score | AI-rated 0-100 score per call, averaged | Measures how well the agent handled conversations |
| Sentiment | Positive / Neutral / Negative breakdown | Reveals overall caller satisfaction |
| Goodbye Rate | Percentage of calls that end with a natural, polite conclusion (the caller says goodbye or thanks) rather than an abrupt hang-up | Indicates whether callers are satisfied or hanging up frustrated |
| Avg. Duration | Mean call length in seconds | Too short = not engaging; too long = agent may be struggling |
| Peak Hours | Hourly call distribution chart | Helps clients staff around AI availability or adjust hours |
Tip: The Goodbye Rate is your single best quality indicator. A high goodbye rate (80%+) means callers are staying on the line, getting their questions answered, and ending the call politely. A low goodbye rate means callers are hanging up ā and that's a signal to improve the agent's greeting, knowledge base, or FAQ answers.
Per-Call Intelligence
Every single call gets its own intelligence report, generated automatically after the call ends. Click any call in the history to see:
A concise 2-3 sentence summary of the entire conversation. Example: "Caller asked about emergency plumbing rates for a burst pipe at a residential address. Agent provided the after-hours rate ($150 call-out fee) and offered to schedule a same-day appointment. Caller declined and said they'd call back in the morning."
This saves your clients from listening to every recording. They can scan summaries in seconds to know exactly what happened.
The system identifies why the caller reached out. Common intents include:
- Appointment booking ā wanted to schedule a visit
- Pricing inquiry ā asked about costs or rates
- Service information ā wanted to know what services are offered
- Hours/location ā asked when or where the business operates
- Complaint ā had an issue to report
- Follow-up ā calling about an existing appointment or service
Over time, intent data reveals what callers care about most ā powerful intelligence for your client's business.
Any information the caller shared is extracted and displayed as structured data:
- Full name
- Phone number
- Email address
- Address or location
- Service requested
- Preferred date/time
This data feeds into the contact record automatically, so your client never has to manually enter caller details.
Each call receives a sentiment score: Positive, Neutral, or Negative. The system analyzes the caller's tone, word choice, and overall conversation trajectory to determine sentiment.
Negative sentiment calls are flagged so your client can follow up personally if needed ā for example, a frustrated caller who wasn't satisfied with the AI's response.
The quality score (0-100) is computed based on multiple factors:
- Did the agent understand the caller's request?
- Was the response accurate and helpful?
- Did the conversation flow naturally?
- Was the call resolved, or did the caller hang up?
- Were any actions taken (booking, message, etc.)?
A score above 80 is excellent. Between 60-80 is acceptable. Below 60 means the agent likely struggled ā check the transcript and consider updating FAQs or the knowledge base.
Contact Intelligence
The platform automatically builds a contact record for every unique caller. Over time, each contact accumulates:
- Full call history with that contact
- Total number of interactions
- Average sentiment across all their calls
- Key data points (name, email, service interests)
- First and last contact dates
This turns your client's AI agent into more than a receptionist ā it becomes a lightweight CRM that builds itself with every call.
Note: Use analytics data in your monthly client reports. Showing a client that their agent handled 247 calls this month with a 92% quality score and 85% positive sentiment is the best retention tool you have. Numbers don't lie ā and they justify your monthly fee.
Using Analytics to Improve Performance
Analytics aren't just for reporting ā they're your roadmap for making the agent better:
| Signal | Likely Issue | Fix |
|---|---|---|
| Low goodbye rate | Callers are frustrated or confused | Shorten the greeting, improve FAQ answers |
| Low quality scores | Agent doesn't have enough information | Add FAQs, upload knowledge base documents |
| High negative sentiment | Agent may be giving wrong info | Review transcripts, correct business profile |
| Very short calls (<15s) | Callers hanging up on greeting | Make greeting shorter and more natural |
| Very long calls (>5min) | Agent going in circles | Add clearer FAQs for common topics |
Was this page helpful?