Lead Source

The “Lead_Source” tracker detects phrases that indicate how a customer first heard about the company, product, or service. By capturing these references, teams can gain insights into which marketing efforts are working, where prospects are coming from, and where to focus future investments.

Who Benefits:

• Project Managers & Team Leads: Understand where leads originate, guiding resource allocation, training focus, and follow-up strategies.
• Sales & CX Teams: Quickly identify common acquisition channels mentioned by customers, enabling personalized engagement and rapport-building.
• Business Managers & Strategists: Spot trends in lead sources, informing marketing spend, campaign planning, and channel optimization.
• Compliance Officers: Ensure that any required disclosures about referral programs or partner relationships are mentioned as needed.

Value Proposition:

• Data-Driven Marketing: Highlighting lead sources supports data-driven decisions on marketing campaigns and budget allocations.
• Improved Targeting: By knowing which channels yield high-quality leads, teams can refine messaging and focus on productive avenues.
• Continuous Improvement: Insights into origin references guide ongoing tweaks to outreach, partnerships, and brand-building efforts.

Data Dictionary

Objective:

Detect phrases in the conversation that indicate how the caller discovered or learned about the company, product, or service. The tracker returns occurrences of matched phrases, timestamps, and confidence, enabling teams to understand lead origins.

Proposed JSON Schema:

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

Data Dictionary:

If No Matches Found:
If no lead source mentions are detected, return an empty matches array for "Lead_Source" or omit the tracker. If no trackers match at all, return:

{
  "trackers": []
}

Determining Confidence & Thresholds

How the Lead_Source Tracker Works:

• Seed the tracker with phrases related to discovering the company (e.g., "I found you on Google," "My friend recommended you," "I saw your booth at the trade show").
• The system interprets these seed phrases and identifies semantically similar statements in the transcript.
• Confidence indicates how closely the detected phrase aligns with indicating a lead source.

Confidence & Calibration:

• Initial Testing: Validate on known calls to ensure detected phrases accurately represent lead sources.
• Refine Phrases: Add or remove seed phrases to capture evolving marketing channels and discovery patterns.
• Thresholding: Downstream consumers can filter out low-confidence matches if needed.
• Stakeholder Feedback: Integrate input from marketing or sales leads to refine which channels matter most.

Prompt Construction & Instructions

Role Specification & Reiteration:

• Present the system as a “highly experienced assistant” identifying lead source mentions.
• Reiterate instructions: return only JSON, no extra commentary, follow schema.

No Hallucination:

• Only detect lead source mentions if supported by transcript evidence.
• If uncertain, assign a 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 'Lead_Source' mentions in a conversation. You have:

• A transcript
• A 'Lead_Source' tracker defined by seed phrases indicating how the customer found the company or product.

Your task:

1. Detect phrases where the customer mentions how they discovered the business.
2. Return text, timestamp, and confidence.
3. Return only JSON per the schema, no extra commentary.

Instructions & Variables (example):

{
  "tracker": {
    "name": "Lead_Source",
    "phrases": [
      "I found you on Google",
      "a friend recommended you",
      "saw your ad on",
      "learned about you at the event"
    ]
  }
}

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

[TRANSCRIPT_JSON][TRANSCRIPT_JSON]"

Expected Output Format:

{
  "trackers": [
    {
      "name": "Lead_Source",
      "matches": [
        {
          "text": "I discovered your product through a colleague’s recommendation.",
          "timestamp": "2024-07-10T10:25:00Z",
          "confidence": 0.9
        }
      ]
    }
  ]
}