Repeat Inquiry

The “Repeat_Inquiry” tracker detects cues that a customer has previously reached out and is following up on an unresolved problem. By capturing these moments, teams can measure how effectively their first-contact resolution strategies work and where improvements are needed to minimize repetitive interactions.

Who Benefits:

• Project Managers & Team Leads: Identify recurring issues that aren’t being resolved promptly, guiding process improvements and training.
• Sales & CX Teams: Spot patterns in unresolved problems, offering opportunities to deliver better solutions or provide additional resources.
• Business Managers & Strategists: Understand which topics or issues frequently lead to repeat inquiries, informing product enhancements or policy changes.
• Compliance Officers: Confirm that repeated requests are handled consistently and according to guidelines, maintaining transparency and trust.

Value Proposition:

• Efficiency Gains: Understanding repeat inquiries helps teams address root causes, reduce repeated calls, and improve customer satisfaction.
• Actionable Insights: Structured JSON output integrates with dashboards, enabling trend analysis and prioritization of systemic fixes.
• Continuous Improvement: By recognizing recurring contact patterns, organizations can refine scripts, policies, and service flows for better outcomes.

Data Dictionary

Objective:

Detect phrases or context indicating that the customer has contacted the service before about the same issue. The tracker returns occurrences of matched phrases, along with timestamps and confidence, helping teams identify when existing resolution strategies fall short.

Proposed JSON Schema:

{
  "trackers": [
    {
      "name": "Repeat_Inquiry",
      "matches": [
        {
          "text": "<string>",
          "timestamp": "<ISO 8601 or relative time>",
          "confidence": <number> // 0.0 to 1.0
        }
      ]
    }
  ]
}

Data Dictionary:

If No Matches Found:
If no repeat inquiries are detected, return an empty matches array for the "Repeat_Inquiry" tracker or omit the tracker. If no trackers match at all, return:

{
  "trackers": []
}

Determining Confidence & Thresholds

How the Repeat_Inquiry Tracker Works:

• Seed the tracker with phrases suggesting the customer has contacted before (e.g., "I reached out last time," "I already called about this," "I’ve been trying to get this resolved for days").
• The system generalizes these phrases to find semantically similar statements.
• Confidence indicates how clearly the phrase represents a repeat inquiry about the same issue.

Confidence & Calibration:

• Initial Testing: Validate on known calls to ensure detected phrases truly indicate a repeat inquiry.
• Refine Phrases: Add or remove seed phrases to improve coverage and reduce false positives.
• Thresholding: Downstream consumers can ignore low-confidence matches if necessary.
• Stakeholder Feedback: Integrate insights from CX leads or compliance teams to refine what qualifies as a repeat inquiry.

Prompt Construction & Instructions

Role Specification & Reiteration:

• Present the system as a “highly experienced assistant” identifying repeat inquiries.
• Reiterate instructions: output only JSON, no extra commentary, follow schema.

No Hallucination:

• Only detect repeat inquiries if supported by transcript evidence.
• If uncertain, assign lower confidence or omit matches.

Strict Formatting:

• Return only JSON with trackers.
• If none found, return empty trackers.

Example Prompt for Implementation

System Message (Role: System):
"You are a highly experienced assistant that identifies 'Repeat_Inquiry' mentions in a conversation. You have:

• A transcript
• A 'Repeat_Inquiry' tracker defined by seed phrases indicating the customer has contacted before about the same issue.

Your task:

1. Detect phrases suggesting the customer is following up on an unresolved problem.
2. Return text, timestamp, and confidence for each match.
3. Return only JSON as per the schema, no extra commentary.

Instructions & Variables (example):

{
  "tracker": {
    "name": "Repeat_Inquiry",
    "phrases": [
      "I called last time",
      "we discussed this issue before",
      "I already reached out",
      "trying to get this resolved again"
    ]
  }
}

User Message (Role: User):
"Analyze the following transcript and detect 'Repeat_Inquiry' mentions:

[TRANSCRIPT_JSON][TRANSCRIPT_JSON]"

Expected Output Format:

{
  "trackers": [
    {
      "name": "Repeat_Inquiry",
      "matches": [
        {
          "text": "I already reached out about this problem yesterday.",
          "timestamp": "2024-07-10T12:20:00Z",
          "confidence": 0.9
        }
      ]
    }
  ]
}