Talk is Data: Mastering Conversations Intelligence

Last updated: June 5, 2026

Every call your AI agent makes is packed with valuable data, but you shouldn't have to listen to every recording to find the "aha!" moments. That’s where Conversation Intelligence (also known as Interaction Evaluations) comes in.

Think of this as a custom "checklist" you give the AI. After every call, the AI reviews the transcript and answers specific questions you’ve defined. It turns raw conversation into structured, actionable data for your reports. 📊

The "Choose-Your-Own-Metrics" Setup

The beauty of this feature is that it’s completely customizable. You aren't stuck with generic "call successful" or "call failed" labels. Instead, you define exactly what you care about. You set the parameters, and the AI handles the rest.

Common Questions You Can Ask the AI:

  • "What was the primary pain point the lead mentioned?"

  • "Did the lead express concern about shipping costs?"

  • "On a scale of 1-10, how interested was the lead in a follow-up?"

  • "Did the lead realize they were talking to an AI?"

How to Build Your Evaluation Fields

Setting up your intelligence metrics is easy. Navigate to the Interaction Evaluations section under your Campaign settings and click on "Create Interaction Evaluation Field." You’ll have two ways to get started:

1. Start from Scratch

Define your own parameters, field types, and descriptions from the ground up. This is perfect for highly specific business requirements.

2. Use a Prebuilt Template

We’ve created four standard templates to help you get moving quickly:

Field Type

Best For...

Example

Customer Feedback (Yes/No)

Tracking specific binary outcomes or mentions.

"Did the lead mention a competitor? (Yes/No)"

Customer Feedback (Text)

Capturing qualitative data or summaries.

"Summarize the lead's main objection in one sentence."

Evaluate Success (Choices)

Grouping calls into predefined categories.

"Lead status: Interested, Not Interested, or Follow-up Required."

Evaluate Success (Number)

Quantifying sentiment or lead quality.

"Rate the lead's purchase intent from 1 to 10."

Pro-Tips for Success 💡

  • Be Specific: In the "Description" box for your field, tell the AI exactly what to look for. The better the instructions, the more accurate the data.

  • Start Small: Begin with 3-4 key metrics (like "Pain Point" and "Budget Mentioned") before diving into more complex evaluations.

  • Check the Insights: Once your calls start rolling in, these fields will be available in your insights section, allowing you to filter and sort your leads based on their actual responses.