Skip to content

CSAT (Customer Satisfaction Score)

CSAT is an AI insight that estimates how satisfied the customer is with the outcome of a conversation. In MiaRec, CSAT is typically represented as:

  • a numeric score (commonly 1–5)
  • optional thresholds/labels (e.g., “Dissatisfied”)
  • a short explanation to help human reviewers trust and act on the score

Where CSAT appears

Once enabled and populated, CSAT is available in:

  • Conversation Details → Analytics (score + label + explanation)
  • Dashboards (average, distribution buckets, clickable drilldowns)
  • Search (e.g., CSAT < 3)

What gets configured

Depending on your deployment, CSAT is configured using:

  1. Custom Field (numeric) – stores the CSAT value and powers dashboards/search
    Menu: Administration > Customization > Custom Fields

  2. AI Task (CSAT scoring) – extracts CSAT from transcript/thread and writes to the field
    Menu: Administration > Speech Analytics > AI Assistant > AI Tasks

  3. Optional Prompt/Filter override – to align scoring to your business definitions

In many deployments, CSAT is provided as a prebuilt AI Task and may already have a prebuilt Custom Field. In that case, you primarily enable the task and validate results.

Step 1 — Create or verify the CSAT Custom Field

Menu: Administration > Customization > Custom Fields

  • Name: CSAT
  • Type: Number (integer)
  • Range: 1–5 (optional: allow 0 = Unknown)
  • Display group: CX Metrics (or similar)
  • AI Insights eligibility: enabled (so tasks can map to this field)
  • Dashboard: enabled (recommended)
  • aggregate: Average

Thresholds (labels + colors)

Recommended thresholds:

Score Label
1 Very Dissatisfied
2 Dissatisfied
3 Neutral
4 Satisfied
5 Very Satisfied

These thresholds create the dashboard buckets that users can click to drill down into matching conversations.

If you don’t see CSAT in Custom Fields

  • In SaaS deployments, fields may be pre-configured by MiaRec.
  • Contact your provider to enable the CSAT field for your tenant, or ask for permission to manage Custom Fields.

Step 2 — Enable the CSAT AI Task

Menu: Administration > Speech Analytics > AI Assistant > AI Tasks

  1. Open the Disabled tab.
  2. Find the CSAT task (name may vary: “CSAT”, “CSAT Scoring”).
  3. Click Enable.

Validate mapping

Open the enabled task and verify the mapping includes: - output attribute (e.g., csat) → Custom Field (CSAT)

Step 3 — (Optional) Override CSAT prompt and/or filters

Menu: Administration > Speech Analytics > AI Assistant > AI Tasks → Enabled → task → Edit

When to override

Override the prompt when: - you want stricter definitions of “resolved” - you want to weigh tone vs outcome differently - you want more consistent “unknown” handling - you need a more helpful explanation for coaching

Override filters when: - you want to score only certain call types (e.g., inbound support calls) - you want a minimum duration threshold (e.g., > 15 seconds)

  • Call duration > 15 seconds (or > 30 seconds if you have many short calls)
  • Inbound calls only (if CSAT is support-focused)

MiaRec prompts commonly extract: - the CSAT score - plus an explanation

Suggested JSON format

{
  "csat": {
    "value": 4,
    "explanation": "The issue was resolved and the customer expressed appreciation before ending the call."
  }
}

Suggested rubric (1–5)

  • 1 = Very Dissatisfied: angry/frustrated, unresolved, negative outcome
  • 2 = Dissatisfied: mostly negative, unresolved or poor handling
  • 3 = Neutral: mixed/unclear, partial resolution
  • 4 = Satisfied: resolved, generally positive
  • 5 = Very Satisfied: praise/thanks, excellent outcome

MiaRec stores explanations automatically alongside the insight value. The explanation is displayed in the Conversation Details → Analytics view next to the CSAT score without requiring a separate custom field.

Step 5 — Test and validate

Use Playground or “Save and Test”:

  1. Test on at least 10 conversations:
  2. 3 clearly positive
  3. 3 clearly negative
  4. 4 mixed/ambiguous
  5. Confirm:
  6. output is valid JSON (if configured)
  7. values are within 1–5
  8. explanations are short and evidence-based

Dashboards

CSAT Dashboard

Figure: CSAT Dashboard showing score distribution, trend over time, and per-user breakdown.

  • Open the CSAT dashboard
  • Confirm bucket counts and clickable drilldowns work as expected

Use advanced search to filter: - CSAT < 3 (dissatisfied calls) - CSAT = 1 (urgent review)

Troubleshooting

CSAT not showing up - transcript missing - task not enabled - filters exclude the conversation - mapping points to a different field

Dashboard empty - metric field exists but isn’t being populated yet - time range too narrow - no backfill of historical data