Skip to content

Conversation Analytics – Administration Guide

This guide explains how tenant administrators configure MiaRec Conversation Analytics for their organization: enabling transcription, enabling AI insights, validating results, and rolling out dashboards and workflows.

New to Conversation Analytics?
Start with Conversation Analytics – Overview & Key Concepts to learn the core concepts (Conversations, Transcripts/Threads, Custom Fields, AI Tasks, Filters, Explanations).

Who this guide is for

  • Tenant administrators of MiaRec SaaS customers
  • Tenant administrators of partner-hosted deployments

What you can configure (tenant view)

As a tenant admin, you typically control:

  • Transcription enablement and validation (for voice calls)
    Note: in some SaaS deployments, transcription engines are managed by MiaRec; tenant admins validate results and request changes.
  • Custom Fields used to store AI insights (scores, categories, extracted values)
  • AI Tasks:
  • enable/disable prebuilt tasks
  • configure or create tenant-specific tasks
  • override prompts and filters (where permitted)
  • Testing and validation using Playground / “Save and Test”
  • Dashboards, search filters, and reporting to make insights actionable
  • Permissions/visibility (role access and field visibility where supported)

What this guide does not cover

  • System/global configuration of engines, global tasks, background jobs, or multi-tenant operations
    → See Conversation Analytics – Platform Setup & Operations Guide (partners/service providers).

How this guide is organized

  1. Core Configuration – transcription, fields, tasks, overrides, Auto QA
  2. Testing and Tuning – Playground + prompt best practices
  3. Dashboards and Search – how metrics become dashboards and filters
  4. Monitoring and Troubleshooting – common issues and fixes
  5. Insight Catalog – step-by-step recipes for specific insights (CSAT, sentiment, topics, etc.)

If you’re implementing Conversation Analytics for the first time, follow this sequence:

  1. Enable and validate transcription (voice calls)
  2. Enable 2–3 prebuilt AI Tasks (e.g., Summarization + Sentiment + CSAT)
  3. Test results on real conversations using Playground
  4. Confirm results appear in Conversation Details, Dashboards, and Search
  5. Roll out dashboards and saved searches to supervisors and QA teams
  6. Expand with custom insights and Auto QA as needed