Lesson 2

Introduction to Analytics

TABLE OF CONTENTS▼

    “Call Analytics” can cover campaigns, individual calls, batch calls, and conversational pathway driven calls.

    Analytics can be found in your main menu:

    Call Analytics (Home)

    You can use “Call Analytics” to gain insights into the results of your calls (how many were completed, transferred, or went to voicemail).   

    Call Analysis

    This screen shows a summary of the call analysis that’s been performed. These results are based on the analysis schema you’ve defined for your calls, either in “Batch Calls” or in “Inbound Phone Numbers.”

    Here is a basic example of analysis (3 questions) performed on a single call.

    Real Time Logs

    Real time logs show you live reports from all agents currently active. This is a streaming log function with several filters available.

    Category

    Call: Logs in the “Call” category are triggered by and related to agents running live calls. 

    Queue: Logs in the “Queue” category are related to the creation and queuing of calls (batch calls, for example). Queue logs are created before a call is performed by an available agent.

    Status

    Info: Informational logs give you updates on call status and progress within conversational pathways.


    Performance: These logs are related to latency and timing. They highlight anything related to the backend of your calls. 


    Error: If there are problems with variables, faulty API webhooks, or a pathway has an issue completing, the logs will appear with an “Error” badge.

    Live Calls Table

    All calls that are currently occurring will be shown in this table. 

    Latency Charts

    The “Latency” tab can be used to ensure that the models (LLMs) that power your agents are performing well in comparison to a human conversation.   

    Audio Latency

    Audio latency refers to the “text to speech” operation that your Bland agent is performing behind the scenes. Increase in this value, or high numbers overall, can indicate an issue with your model selection or prompt design. 

    Text Latency

    Text latency refers to the time it takes to send prompts/get responses back from your agent’s models. Problems with this latency could indicate that there are issues with model performance problems or that too much data/prompt information has been provided. 

    Agent Response Time Chart

    This chart helps you understand how quickly your agent is responding across conversations.

    Ideal Latency Ranges

    Metric

    Target Range

    Description

    Audio Latency

    < 200 ms

    Human auditory processing speed is around this range

    Text Latency

    400-800ms

    Internal Bland latency sending prompts to the LLMs and receiving responses

    Total Latency (Turnaround Time)

    500 - 1000ms

    Natural conversational pauses and human-like processing and response times

    NEXT

    Feedback and Monitoring

    Feedback and Monitoring

    NEXT

    Feedback and Monitoring

    Table of contents▼
      Scroll to top