Sales Question

The "Sales Question" tracker detects instances where a customer inquires about buying a product, exploring pricing, seeking discounts, or understanding payment terms. Recognizing these inquiries helps teams measure customer interest, evaluate agent responses, and improve pricing strategies or promotional messaging.

Who Benefits:

• Project Managers & Team Leads: Understand common pricing or discount queries, guiding training on providing transparent and helpful responses.
• Sales & CX Teams: Quickly identify when customers show buying interest, enabling real-time or post-call coaching and interventions.
• Business Managers & Strategists: Spot trends in what customers ask before purchasing, informing product development, pricing models, and promotional strategies.
• Compliance Officers: Ensure agents respond to pricing and discount questions in accordance with company policy and regulatory standards.

Value Proposition:

• Informed Strategies: Detecting sales-related questions highlights gaps in current offerings or communication, aiding strategic improvements.
• Operational Efficiency: Machine-readable JSON output integrates with dashboards, supporting data-driven decisions about products and pricing.
• Continuous Improvement: Insights from customer queries guide refinements to scripts, self-service tools, or marketing materials.

Data Dictionary

Objective:

Detect phrases where a customer asks about buying conditions, discounts, or pricing. The tracker returns occurrences of matched phrases, along with timestamps and confidence. This data helps teams understand what drives customer interest and how often customers inquire about sales conditions.

Proposed JSON Schema:

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

Data Dictionary:

If No Matches Found:
If no sales-related questions are detected, return an empty matches array for "Sales_Question" or omit the tracker. If no trackers match at all, return:

{
  "trackers": []
}

Determining Confidence & Thresholds

How the Sales_Question Tracker Works:

• Define seed phrases that represent interest in buying or discounts (e.g., "Can I get a discount?", "What’s the price?", "Do you offer a payment plan?").
• The system interprets these phrases and identifies similar statements.
• Confidence reflects how closely the detected phrase aligns with sales-related questions.

Confidence & Calibration:

• Initial Testing: Validate on known calls to confirm detected phrases accurately represent sales inquiries.
• Refine Phrases: Add or remove seed phrases to capture evolving language patterns or new promotions.
• Thresholding: Downstream processes can filter low-confidence matches.
• Stakeholder Feedback: Involve sales leads, CX managers, or compliance officers to refine what phrases count as sales inquiries.

Prompt Construction & Instructions

Role Specification & Reiteration:

• Present the system as a “highly experienced assistant” identifying sales-related questions.
• Reiterate instructions: return only JSON, no extra commentary, follow schema.

No Hallucination:

• Only detect sales questions if clearly stated.
• 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 'Sales_Question' mentions in a conversation. You have:

• A transcript
• A 'Sales_Question' tracker defined by seed phrases indicating inquiries about buying, pricing, or discounts.

Your task:

1. Detect phrases where the customer asks about purchasing conditions, discounts, or pricing.
2. Return text, timestamp, and confidence for each match.
3. Return only JSON according to the schema, no extra commentary.

Instructions & Variables (example):

{
  "tracker": {
    "name": "Sales_Question",
    "phrases": [
      "Can I get a discount",
      "What's the price",
      "Do you offer a payment plan",
      "How much does it cost",
      "Any promotions available"
    ]
  }
}

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

[TRANSCRIPT_JSON][TRANSCRIPT_JSON]"

Expected Output Format:

{
  "trackers": [
    {
      "name": "Sales_Question",
      "matches": [
        {
          "text": "Is there any promotion going on right now?",
          "timestamp": "2024-07-10T09:15:00Z",
          "confidence": 0.9
        }
      ]
    }
  ]
}