Contact Information Collection

This guide explains how the system collects and manages contact information across all channels (SMS, Email, Chat, Voice) to create and maintain accurate client records.

Overview

The system automatically collects contact information from conversations to create client records and personalize interactions. Each channel has specific collection requirements and methods.

Channel-Specific Collection

SMS

Required Information:

• Phone Number: Automatically known from sender (channel_identifier)
• Name: Requested after first message if missing

Collection Process:

1. First Message: Customer sends SMS
2. Name Request: If name is missing, agent asks: "Hi! Before we continue, could I get your name?" or "What's your name?"
3. Identity Verification: If phone matches existing client, agent verifies: "I think we've spoken before - is this [FirstName]?"
4. Client Creation: Once name is collected, client record is created/linked
5. Personalization: Agent uses first name throughout conversation

Special Features:

• Known Texter Verification: System checks if phone number matches existing client
• Natural Verification: Agent verifies identity conversationally (not template-like)
• Early Verification: Happens within first 2 messages
• Name Usage: Agent addresses customer by first name once confirmed

Phone Number Priority:

• Sender's phone number (from channel_identifier) always takes priority
• Extracted phone numbers from message content are secondary
• Ensures accurate phone storage for SMS clients

Email

Required Information:

• Email Address: Automatically known from sender (channel_identifier)
• Name: Requested for new conversations if missing

Collection Process:

1. Email Received: System identifies sender email address
2. Automatic Extraction: AI extracts name, phone, company from email content
3. Name Request: For new conversations, agent asks for name if missing
4. Client Creation: Client record created/linked based on email address
5. Personalization: Agent uses name when addressing customer

Extraction Methods:

• AI-assisted contact info extraction: Optional extraction using pattern matching and model support
• Regex Pattern Matching: Basic extraction for workflow-specific needs
• Email Signature Parsing: Extracts information from email signatures

Best Practices:

• Include extract_info step in workflows for additional extraction
• Review automatically created clients regularly
• Merge duplicate clients when found

Chat (Web Widget)

Required Information:

• Name: Required before first message
• Email: Required before first message

Collection Process:

1. Widget Opens: Customer clicks chat widget
2. Information Form: Widget displays form requesting name and email
3. Validation: System validates email format
4. Client Creation: Once both fields collected:
   • Client record created or linked
   • Conversation linked to client
   • Messaging enabled
5. Additional Extraction: Phone, company extracted from messages if mentioned

Enforcement:

• Customers cannot send first message until name and email are provided
• Form validation ensures proper email format
• System prevents messaging until required information is collected

Voice

Required Information:

• Phone Number: Automatically known from caller (channel_identifier)
• Name: Requested during conversation if missing

Collection Process:

1. Call Received: System identifies caller phone number
2. Name Request: Agent asks for name during conversation
3. Client Creation: Client record created/linked based on phone number
4. Personalization: Agent uses name throughout call

Name Extraction and Validation

Extraction Methods

Pattern Matching:

• Explicit introductions: "my name is", "i'm", "i am", "this is", "call me"
• First Last pattern: Two capitalized words
• Negative lookahead: Excludes phrases like "looking for", "trying to"

Validation:

• Validates against false positive list
• Rejects common verbs/nouns (Looking, Trying, Interested, etc.)
• Checks for proper capitalization
• Validates minimum length

False Positive Prevention

Common False Positives Rejected:

• "Looking" from "I'm looking for..."
• "Trying" from "I'm trying to..."
• "Interested" from "I'm interested in..."
• Common greetings: "Hello", "Thank", "Please", etc.

Validation Rules:

• Single-word names checked against false positive list
• Verb forms (ending in -ing, -ed) rejected for short names
• Must start with capital letter
• Minimum length requirements

Examples:

• ✅ Valid: "I'm John Smith" → Extracts "John" and "Smith"
• ✅ Valid: "My name is Sarah" → Extracts "Sarah"
• ❌ Invalid: "I'm looking for a truck" → Does NOT extract "Looking"
• ❌ Invalid: "I'm trying to find..." → Does NOT extract "Trying"

Client Record Creation

Automatic Creation

Clients are automatically created when:

• Contact information is extracted from conversations
• Required information is collected (varies by channel)
• Email or phone number is available

Matching and Merging

Matching Logic:

1. Check existing clients by email (case-insensitive)
2. If no email match, check by phone (normalized comparison)
3. If match found: Update existing client with new information
4. If no match: Create new client record

Phone Number Normalization:

• Removes formatting: dashes, dots, spaces, parentheses
• Normalizes country code (+1 prefix handling)
• Compares normalized versions for matching

Email Matching:

• Case-insensitive comparison
• Exact match required
• Used as primary identifier for email conversations

Information Priority

When Multiple Sources Available:

1. SMS: Sender's phone number > Extracted phone from content
2. Email: Sender's email address > Extracted email from content
3. Name: Client record > Extracted from messages > Contact data
4. Company: Client record > Extracted from messages

Identity Verification

For Known Contacts

SMS Identity Verification:

• System checks if phone number matches existing client
• If found and conversation not linked, agent verifies identity
• Verification happens early (first 2 messages)
• Natural, conversational verification (not template-like)

Verification Examples:

• "I think we've spoken before - is this [FirstName]?"
• "This number looks familiar - are you [FirstName]?"
• "Hey, are you [FirstName]?"

After Verification:

• Conversation linked to client record
• Agent uses client's first name throughout
• All future messages personalized with name

Best Practices

Data Quality

1. Review Regularly: Check automatically created clients for accuracy
2. Validate Names: Ensure extracted names are real (not false positives)
3. Merge Duplicates: Combine duplicate client records when found
4. Complete Information: Add missing details as information becomes available

Channel-Specific

SMS:

• Trust sender's phone number (always accurate)
• Verify identity for known texters naturally
• Use first name once confirmed

Email:

• Sender email is always accurate
• Extract additional info from email content
• Use extract_info workflow step for specific needs

Chat:

• Enforce name and email collection before messaging
• Validate email format
• Create client record immediately upon collection

Error Handling

Invalid Names:

• System automatically rejects false positives
• Review clients with unusual names
• Manually correct if needed

Missing Information:

• System prompts for missing required information
• Agent asks naturally in conversation
• Information collected before proceeding

Duplicate Clients:

• System attempts to match by email/phone
• Manually merge if duplicates created
• Review matching logic if issues persist

Troubleshooting

Names Not Extracting

• Check if name is mentioned explicitly
• Verify name format (capitalized, proper structure)
• Review false positive validation (may be too strict)
• Check PII extraction logs

False Positive Names

• System should automatically reject these
• If false positive appears, review extraction logic
• Manually correct client record
• System learns from corrections

Missing Contact Information

• Verify required fields are being collected
• Check channel-specific requirements
• Review information collection prompts
• Ensure agent is asking for missing information

Clients Not Linking

• Verify contact information matches (email/phone)
• Check normalization logic
• Review matching criteria
• Manually link if needed

Related Documentation

• Clients Overview - Client management basics
• SMS Conversations - SMS contact collection
• Email Workflows - Email data extraction
• Chat Setup - Chat information collection