Skip to main content
Post-Call Metrics let you automatically extract and analyze data from every call. Define what you want to measure, and Atoms analyzes conversations to provide structured insights.
[IMAGE: Post-call metrics setup modal showing form]

​
Location

Left Sidebar → Post Call Metrics

​
What Are Post-Call Metrics?

After each call, Atoms can analyze the conversation and extract specific information:
  • Did the issue get resolved?
  • What product was discussed?
  • How satisfied was the customer?
  • Does this need follow-up?
This data helps you understand performance and improve over time.

​
Set Up Post Call Metrics Modal

​
Left Panel - Metric Types

TypeIconDescription
Disposition Metrics📊Evaluate call outcomes by category
Templates📋Select from pre-filled metrics

​
Right Panel - Configuration

Header shows “Disposition Metrics” with “Add Another +” button.

​
Creating a Metric

​
Metric Form Fields

FieldTypeRequiredDescription
IdentifierTextYesUnique name for this metric
Data TypeDropdownYesString, Number, Boolean
PromptText areaYesQuestion for AI to analyze

​
Identifier Rules

  • Lowercase letters only
  • Numbers allowed
  • Underscores allowed
  • No spaces or special characters
Examples:
  • call_outcome ✓
  • satisfaction_score ✓
  • follow_up_needed ✓
  • Call Outcome ✗ (has spaces, uppercase)

​
Data Types

TypeWhen to UseExample Values
StringCategories, text answers”resolved”, “billing”, “product_x”
NumberScores, counts1-5 rating, count of issues
BooleanYes/no questionstrue, false

​
Example Metrics

​
Call Outcome (String)

Identifier: call_outcome
Data Type: String
Prompt: What was the outcome of this call? 
        Options: resolved, unresolved, transferred, callback_scheduled

​
Satisfaction Score (Number)

Identifier: satisfaction_score
Data Type: Number
Prompt: Based on the conversation tone and customer responses, 
        rate the customer's satisfaction from 1 (very dissatisfied) 
        to 5 (very satisfied).

​
Follow-up Needed (Boolean)

Identifier: follow_up_needed
Data Type: Boolean
Prompt: Does this call require any follow-up action from 
        the support team? Answer true or false.

​
Product Discussed (String)

Identifier: product_discussed
Data Type: String
Prompt: What product or service was primarily discussed 
        in this conversation?

​
Using Templates

Click Templates in the left panel to see pre-built metric sets:
  • Common support metrics
  • Sales qualification metrics
  • Satisfaction metrics
Templates pre-fill identifier, type, and prompt. Customize as needed.

​
Viewing Results

After calls complete, metrics appear in:
  • Conversation Logs — Individual call details
  • Analytics Dashboard — Aggregate views
  • Webhook Payloads — Sent to your systems (Analytics Completed event)

​
Best Practices

​
Write Clear Prompts

The AI uses your prompt to analyze. Be specific: Vague: “Was the customer happy?” Better: “Based on the customer’s language, tone, and explicit statements, rate their satisfaction from 1-5.”

​
Use Consistent Identifiers

Stick to a naming convention across agents for easier analysis.

​
Start Simple

Begin with 3-5 key metrics. You can always add more later.

​
Test with Real Calls

Make test calls and verify metrics are extracted correctly.

​
What’s Next

Testing Your Agent

Test metric extraction

Analytics

View aggregated metrics