AI Chat with Product Listing Ads - Technical Overview

⚠️ Important: This use case demonstrates conversational AI chat interfaces with embedded Product Listing Ads. Users interact with a chat assistant that provides natural language responses with product recommendations displayed as compact horizontal cards within the conversation flow.

Live Demo

A complete working demo is available at https://demo-aura.affinity.net/ai-chat-pla showing real Affinity AI Bid API integration with IAB Native Ads 1.2 format.

Note: The demo uses simplified pattern matching for keyword extraction. Production implementations should use LLM-powered intent extraction.

Executive Summary

This document outlines how publishers can integrate Product Listing Ads into conversational AI chat experiences using standard IAB Native Ads 1.2 format with OpenRTB 2.6. Users ask natural language questions about products, and the system responds with helpful information alongside relevant product ads displayed as compact cards embedded within chat messages.

Key Benefits:

Conversational Discovery: Natural language product search within chat interface
Natural Language Price Filtering: Extract price constraints from queries like "under $200"
Embedded Product Cards: Compact horizontal cards within chat messages, not separate sections
Industry Standard: Uses OpenRTB 2.6 + IAB Native Ads 1.2 specification
Easy Integration: Works with existing ad tech ecosystem (DSPs, SSPs, ad servers)
Streaming Responses: Word-by-word text streaming for responsive UX
Privacy-First: Contextual targeting without PII requirements
Engagement-Based: Revenue from clicks and conversions on product cards

Compare With:

AI Search Product Ads displays product cards above search results
AI Search LLM Weaving weaves sponsored text into LLM responses
Conversational AI provides general conversational monetization patterns

Use Case Example

User Query: "What are the best bluetooth speakers under $200?"

Chat Response with Products:

┌──────────────────────────────────────────────────────────────┐
│  You: What are the best bluetooth speakers under $200?      │
└──────────────────────────────────────────────────────────────┘

┌──────────────────────────────────────────────────────────────┐
│  🤖 Assistant:                                                │
│                                                              │
│  When choosing a bluetooth speaker under $200, consider     │
│  key features like sound quality, battery life, and          │
│  portability. Look for waterproof ratings if you plan to     │
│  use it outdoors. Here are some options available:          │
│                                                              │
│  ┌────────────────────────────────────────────────────┐    │
│  │ [Image] JBL Flip 6          $129.99  [View] →     │    │
│  │         Amazon                                     │    │
│  └────────────────────────────────────────────────────┘    │
│                                                              │
│  ┌────────────────────────────────────────────────────┐    │
│  │ [Image] Sony SRS-XB43       $178.00  [View] →     │    │
│  │         Best Buy                                   │    │
│  └────────────────────────────────────────────────────┘    │
│                                                              │
│  ┌────────────────────────────────────────────────────┐    │
│  │ [Image] Bose SoundLink      $149.99  [View] →     │    │
│  │         Target                                     │    │
│  └────────────────────────────────────────────────────┘    │
│                                                              │
│  Compare features like battery life, audio codecs, and      │
│  connectivity options to find the best fit for your needs.  │
└──────────────────────────────────────────────────────────────┘

User Interaction:

User asks question in natural language with price constraint
System extracts keywords ("bluetooth speakers") and price filter ("under $200")
Assistant response streams word-by-word
Product cards appear embedded within the chat message
Each card shows: image, title, price, merchant, CTA button
Entire card is clickable and navigates to merchant's product page
Impression tracked when card is 50% visible
Click tracked when user clicks on card

Why IAB Native Ads for Chat Products?

1. Industry Standard

Built on OpenRTB 2.6 and IAB Native Ads 1.2 specification
No custom integrations needed for each retailer or brand
Works with existing ad tech ecosystem (DSPs, SSPs, ad servers)
Broad support across the advertising industry

2. Structured Product Data

Retailers provide structured product data via native assets
Products displayed as compact cards within chat messages
Standard asset types ensure consistency and quality
Easy to render in any chat interface

Custom affinity_ai_product_search extension for price filtering
Extract constraints from natural language
Filter products at bid request time
Improves relevance and user satisfaction

4. LLM-Driven Intent Extraction

Your LLM intelligently extracts user intent from natural language queries
Identifies product keywords and price constraints automatically
Calls Affinity AI Bid API with extracted parameters
Generates helpful response text independent of auction results

5. Privacy-First Design

No PII required in bid requests
Contextual targeting based on chat query and keywords
GDPR/CCPA compliant by design

6. Rich Analytics

Track impressions (when product card becomes visible)
Track clicks (when users click on product cards)
Track conversions (when users make purchases)
Standard tracking mechanisms across all publishers

7. Quality Control

IAB Native specification defines asset requirements
Validation ensures quality before serving
Affinity AI Bid API returns relevant products based on keywords and price filter

Architecture Overview

Key Points:

LLM extracts intent - Your AI intelligently extracts keywords and price constraints from natural language
LLM calls Affinity AI Bid API - Uses extracted parameters to request relevant product ads via OpenRTB
Affinity AI Bid API returns ads - Standard OpenRTB auction returns product listings in IAB Native 1.2 format
LLM generates response - Creates helpful text response independent of the specific products returned
Chat interface renders - Displays LLM text response with embedded product cards
Standard tracking - Impression and click tracking via OpenRTB

Natural Language Price Filtering

Supported Price Patterns

The system can extract price constraints from natural language queries:

Pattern Types:

Pattern	Example Queries	Extracted Filter
Under	"under $200", "below$ 150"	`max_price: 200`
Over	"over $50", "above$ 100"	`min_price: 100`
Between	"between $100 and$ 200"	`min_price: 100, max: 200`
Around	"around $150" (±20% range)	`min: 120, max: 180`
Less than	"less than $100"	`max_price: 100`
At least	"at least $75"	`min_price: 75`
Range	" $100 to$ 200"	`min: 100, max: 200`

The extracted filter is sent to Affinity AI Bid API via the affinity_ai_product_search extension:

{
  "imp": [
    {
      "id": "imp-1",
      "native": {...},
      "ext": {
        "affinity_ai_product_search": {
          "max_price": 200.0,
          "price_currency": "USD"
        }
      }
    }
  ]
}

Benefits:

Improves product relevance by filtering at auction time
Reduces unnecessary product returns
Better user experience with price-appropriate products
Supports any currency (USD, EUR, GBP, etc.)

IAB Native Ads Asset Mapping

Product Listing Native Assets

Following IAB Native Ads 1.2 specification, chat product listings use these standard asset types:

Required Assets:

IAB Asset Type	ID	Data Type	Purpose
Title	1	-	Product name
Image (Main)	2	type: 3	Product image
Data (Price)	3	type: 3	Product price

Optional Assets:

IAB Asset Type	ID	Data Type	Purpose
Data (Desc)	4	type: 2	Product description
Data (Sponsor)	5	type: 1	Merchant/brand name
Data (CTA)	6	type: 12	Call-to-action text

Note: The implementation uses flexible asset mapping that checks both asset IDs and data types for compatibility with different bidders.

Native Request Structure

The native request follows IAB Native Ads 1.2 specification:

{
  "ver": "1.2",
  "assets": [
    {
      "id": 1,
      "required": 1,
      "title": {
        "len": 100
      }
    },
    {
      "id": 2,
      "required": 1,
      "img": {
        "type": 3,
        "w": 300,
        "h": 250
      }
    },
    {
      "id": 3,
      "required": 1,
      "data": {
        "type": 3,
        "len": 50
      }
    },
    {
      "id": 4,
      "required": 0,
      "data": {
        "type": 2,
        "len": 200
      }
    },
    {
      "id": 5,
      "required": 0,
      "data": {
        "type": 1,
        "len": 50
      }
    },
    {
      "id": 6,
      "required": 0,
      "data": {
        "type": 12,
        "len": 30
      }
    }
  ]
}

Key Points:

Asset ID 1: Product title (required, max 100 chars)
Asset ID 2: Product image (required, 300x250px, type 3 = main image)
Asset ID 3: Price (required, max 50 chars, data type 3 = price)
Asset ID 4: Description (optional, max 200 chars, data type 2 = description)
Asset ID 5: Merchant name (optional, max 50 chars, data type 1 = sponsored by)
Asset ID 6: CTA text (optional, max 30 chars, data type 12 = CTA)

OpenRTB Integration

Note: We use standard OpenRTB 2.6 with IAB Native Ads 1.2 for product listings with the affinity_ai_product_search extension for price filtering.

Bid Request Structure

Your chat platform sends standard OpenRTB 2.6 bid requests with native ad placement and price filter:

{
  "id": "ai-chat-pla-req-1704123456789",
  "imp": [
    {
      "id": "imp-1",
      "native": {
        "request": "{\"ver\":\"1.2\",\"assets\":[{\"id\":1,\"required\":1,\"title\":{\"len\":100}},{\"id\":2,\"required\":1,\"img\":{\"type\":3,\"w\":300,\"h\":250}},{\"id\":3,\"required\":1,\"data\":{\"type\":3,\"len\":50}},{\"id\":4,\"required\":0,\"data\":{\"type\":2,\"len\":200}},{\"id\":5,\"required\":0,\"data\":{\"type\":1,\"len\":50}},{\"id\":6,\"required\":0,\"data\":{\"type\":12,\"len\":30}}]}",
        "ver": "1.2"
      },
      "ext": {
        "affinity_ai_product_search": {
          "max_price": 200.0,
          "price_currency": "USD"
        }
      }
    },
    {
      "id": "imp-2",
      "native": {...},
      "ext": {
        "affinity_ai_product_search": {
          "max_price": 200.0,
          "price_currency": "USD"
        }
      }
    },
    {
      "id": "imp-3",
      "native": {...},
      "ext": {
        "affinity_ai_product_search": {
          "max_price": 200.0,
          "price_currency": "USD"
        }
      }
    }
  ],
  "site": {
    "id": "ai-chat-pla",
    "keywords": "bluetooth speaker,wireless speaker"
  },
  "device": {
    "language": "en",
    "geo": {
      "country": "US"
    }
  }
}

Key Request Components:

Request ID: Unique identifier for the bid request
Impressions: Array of impression objects (typically 3 for chat product cards)
Native Request: JSON-encoded IAB Native Ads 1.2 specification
Price Filter Extension: affinity_ai_product_search with price constraints
Site Keywords: Extracted keywords from user query for contextual targeting
Device: User device and geo information for targeting

Determining Number of Product Cards

Publishers control how many product cards to request and display:

Best Practices:

Start with 3 product cards for most queries
Consider query complexity and screen size
Mobile: 2-3 cards work well
Desktop: 3-5 cards can be displayed
Monitor engagement rates and adjust

Bid Response Structure

Retailers respond with standard IAB Native Ads 1.2 format:

{
  "id": "ai-chat-pla-req-1704123456789",
  "seatbid": [
    {
      "seat": "affilizz-products",
      "bid": [
        {
          "id": "bid-001",
          "impid": "imp-1",
          "price": 1.8,
          "adm": "{\"native\":{\"ver\":\"1.2\",\"link\":{\"url\":\"https://amazon.com/product?ref=aura\",\"clicktrackers\":[\"https://track.example.com/click?id=123\"]},\"assets\":[{\"id\":1,\"title\":{\"text\":\"JBL Flip 6 Bluetooth Speaker\"}},{\"id\":2,\"img\":{\"url\":\"https://cdn.example.com/jbl-flip-6.jpg\",\"w\":300,\"h\":250}},{\"id\":3,\"data\":{\"type\":3,\"value\":\"129.99 USD\"}},{\"id\":4,\"data\":{\"type\":2,\"value\":\"Portable waterproof speaker with 12 hours of playtime\"}},{\"id\":5,\"data\":{\"type\":1,\"value\":\"Amazon\"}},{\"id\":6,\"data\":{\"type\":12,\"value\":\"View Product\"}}],\"imptrackers\":[\"https://track.example.com/imp?id=123\"]}}",
          "adomain": ["amazon.com"],
          "cid": "campaign-123",
          "crid": "creative-456",
          "cat": ["IAB19-6"],
          "nurl": "https://track.example.com/win?price=${AUCTION_PRICE}"
        }
      ]
    }
  ],
  "cur": "USD"
}

Native Response Breakdown:

The adm field contains a JSON-encoded IAB Native Ads 1.2 response:

Link: Product URL with optional click trackers
Assets:
- Asset 1 (Title): Product name - "JBL Flip 6 Bluetooth Speaker"
- Asset 2 (Image): Product image URL (300x250px)
- Asset 3 (Data/Price): Price with currency - "129.99 USD" (data type 3)
- Asset 4 (Data/Description): Product description (data type 2)
- Asset 5 (Data/Sponsored): Merchant name - "Amazon" (data type 1)
- Asset 6 (Data/CTA): Call-to-action text - "View Product" (data type 12)
Impression Trackers: Array of impression tracking URLs
Win Notice (nurl): URL with ${AUCTION_PRICE} macro for win notification

Important: The native response can be in the adm field directly or wrapped in a native object for legacy compatibility. The parser handles both formats.

Product Display Integration

Important: Products Are Embedded in Chat Messages

Product cards are displayed as compact horizontal cards WITHIN chat messages, NOT as separate sections. Your platform receives product data from Affinity AI and renders them as embedded cards in the assistant's response message.

Parsing Native Ad Response

Your platform extracts product information from the IAB Native Ads 1.2 assets:

Parse the JSON-encoded adm field from the bid response
Handle both wrapped ({native: {...}}) and unwrapped formats
Extract native assets by ID and data type:
- Asset ID 1 (Title) → Product name
- Asset ID 2 (Image) → Product image URL
- Asset ID 3 (Data type 3) → Price with currency
- Asset ID 4 (Data type 2) → Product description (optional)
- Asset ID 5 (Data type 1) → Merchant name (optional)
- Asset ID 6 (Data type 12) → Call-to-action text (optional, defaults to "View Product")
- Link URL → Product page URL
- Impression trackers → Array of tracking URLs
- Click trackers → Array of tracking URLs
Validate all URLs are HTTPS
Sanitize text content to prevent XSS
Render products as compact horizontal cards in chat message

Implementation Note: Use flexible asset mapping that checks both asset IDs and data types for compatibility with different bidders.

Rendering Product Cards

Display each product as a compact horizontal card embedded in chat message:

Horizontal Layout: Image on left, details on right
Compact Size: Fits naturally within chat flow
Product Image: From Asset ID 2 (fixed height, auto width)
Product Title: From Asset ID 1 (1-2 line clamp for overflow)
Price: From Asset ID 3 (bold, prominent display)
Merchant: From Asset ID 5 (small text)
CTA Button: From Asset ID 6, defaults to "View Product"
Clickable Card: Entire card is clickable, navigates to product URL

Example Product Card Layout:

┌────────────────────────────────────────────────┐
│ [Image]  JBL Flip 6 Bluetooth Speaker         │
│ 80x60px  $129.99                               │
│          Amazon                [View Product →]│
└────────────────────────────────────────────────┘

Responsive Design:

Desktop: Cards stack vertically with 8px gap
Mobile: Same layout, optimized for touch
Hover State: Subtle elevation effect on desktop

Accessibility Features:

WCAG 2.1 AA compliant
Keyboard navigation (Tab, Enter)
Screen reader support with ARIA labels
Focus indicators
High contrast mode support

Tracking Implementation

Fire tracking pixels at appropriate times using navigator.sendBeacon() with image pixel fallback:

Impression Trackers: When product card becomes 50% visible in viewport (Intersection Observer)
Click Trackers: When user clicks on product card (max 200ms wait before navigation)
Win Notice (nurl): Immediately after receiving bid response with ${AUCTION_PRICE} macro replaced

Implementation Details:

Uses Intersection Observer API with 0.5 threshold (50% visibility)
Fires impression only once per product per session (tracked in sessionStorage)
Fires impression before click if not already fired
All tracking URLs validated as HTTPS
Graceful degradation if tracking fails

Integration Workflow

1. LLM Extracts User Intent

Your LLM processes the user's natural language query and extracts:

Keywords: Product-related terms for contextual targeting
Price Constraints: Budget limits expressed in natural language
Intent: Determines if this warrants a product search

Example: "best bluetooth speakers under $200"

Keywords: ["bluetooth speaker", "wireless speaker"]
Price Filter: max_price: 200, currency: USD

2. Build OpenRTB Bid Request

Construct a standard OpenRTB 2.6 bid request with:

Native placement (IAB Native Ads 1.2)
Keywords in site.keywords field
Price filter in imp[].ext.affinity_ai_product_search
Multiple impressions (one per product slot)

See Bid Request Structure section for complete example.

3. Call Affinity AI Bid API

Send the bid request to Affinity AI Bid API endpoint. The API will:

Run a real-time auction with product advertisers
Apply the price filter at auction time
Return up to N products (one per impression requested)
Include tracking URLs for impressions and clicks

4. API Returns Product Ads

Affinity AI Bid API responds with:

Native ads in IAB Native Ads 1.2 format
Product assets: title, image, price, merchant, CTA
Tracking URLs: impression trackers, click trackers, win notice
Metadata: auction price, advertiser domain, categories

See Bid Response Structure section for complete example.

5. Render Products in Chat

Your platform renders the products:

Parse native assets from the response
Display products as cards embedded in chat message
Make cards clickable linking to product URLs
Implement tracking (impressions and clicks)

See Product Display Integration section for parsing details.

6. Fire Tracking Pixels

Fire the provided tracking URLs at appropriate times:

Win Notice (nurl): Immediately after receiving bid response
Impression Trackers: When product becomes visible to user
Click Trackers: When user clicks on product

Use navigator.sendBeacon() or image pixel fallback for reliability.

Benefits for Chat Publishers

Revenue Generation

Engagement-Based: Revenue tied to clicks and conversions
High Intent: Product queries indicate strong purchase intent
Premium CPCs: Product listing ads command premium rates
Standard Tracking: Industry-standard impression and click tracking
Natural Integration: Products embedded in conversation flow

User Experience

Conversational: Natural language product discovery
Price-Aware: Respects user's stated budget constraints
Helpful: Educational content alongside product options
Embedded: Products feel like part of the conversation
Streaming: Responsive word-by-word text display

Technical Advantages

Industry Standard: OpenRTB 2.6 + IAB Native Ads 1.2
Easy Integration: Works with existing ad tech infrastructure
Price Filtering: Natural language price constraint extraction
Scalable: Handle millions of chat queries per day
Broad Support: Compatible with all major ad platforms

Analytics & Optimization

Standard Metrics: Track impressions, clicks via IAB mechanisms
A/B Testing: Test different card layouts and response styles
Performance Data: Optimize based on click-through rates
Revenue Reporting: Standard programmatic advertising metrics

Implementation Considerations

Quality Control

Relevance Threshold: Only show highly relevant products
Asset Validation: Ensure all required native assets are present (title, image, price)
HTTPS Validation: All URLs (images, product links, trackers) must be HTTPS
XSS Prevention: Sanitize all text content before rendering
Image Quality: Validate image URLs are accessible
Flexible Parsing: Handle both wrapped and unwrapped native response formats
Performance: Target <50ms per product parsing

Privacy & Compliance

No PII Required: Contextual targeting only
GDPR Compliant: User consent management
CCPA Compliant: Privacy string support
Transparent: Clear presentation as sponsored products
Session Storage: Tracking state limited to session only

Performance

API Timeout: Set appropriate timeout for bid requests (recommended: 500ms for production)
Async Tracking: Fire tracking pixels asynchronously to avoid blocking user interactions
Graceful Degradation: Handle no-bid responses (empty seatbid array) gracefully
Recommended Targets:
- Bid request construction: <10ms
- Bid API call: <500ms timeout
- Product parsing: <50ms per product
- Impression tracking: fire asynchronously, don't block rendering
- Click tracking: fire asynchronously, don't delay navigation

Native Ad Parsing

JSON Parsing: Properly parse JSON-encoded adm field
Format Compatibility: Handle both wrapped ({native: {...}}) and unwrapped formats
Flexible Asset Mapping: Check both asset IDs and data types for compatibility
Error Handling: Handle missing or malformed assets gracefully
Validation: Validate native response structure and required fields
HTTPS Enforcement: Reject non-HTTPS URLs
XSS Prevention: Sanitize all text content
No Fallback: Real API only, no mock data

API Response Handling

No-Bid Scenarios

The API may return no bids in several scenarios:

No products match the keywords and price filter
All products filtered out by price constraints
No advertisers bidding at this time
Query doesn't match any product categories

Response: Empty seatbid array or no seatbid field

{
  "id": "ai-chat-pla-req-1704123456789",
  "seatbid": [],
  "cur": "USD"
}

Handling: Your platform should gracefully handle this case without showing product cards.

Partial Bids

The API may return fewer products than requested impressions:

Requested 3 impressions, received 2 bids
Some impression slots had no matching products

Handling: Display only the products returned, don't show empty placeholders.

Error Responses

If the API returns an error, handle it gracefully with appropriate error logging and user feedback.

AI Search Product Ads - Product cards above search results
AI Search LLM Weaving - Sponsored text in LLM responses
Conversational AI - General conversational monetization
AI Search SEM Text Ads - Text ads in search results

OpenRTB Components

Request Format - OpenRTB 2.6 bid request format
Response Format - OpenRTB 2.6 bid response format
Native Object - Native ad request structure

Context Enhancement

Context Enhancement Overview - Rich context signals for better targeting

External Standards

OpenRTB 2.6 Specification - IAB Tech Lab
IAB Native Ads 1.2 - IAB Tech Lab
IAB Content Taxonomy - IAB Tech Lab

Version History

v2.0 (2026-01-14): Updated based on actual demo implementation
- Complete rewrite to match actual ai-chat-pla feature implementation
- Added natural language price filtering with affinity_ai_product_search extension
- Added OpenRTB 2.6 + IAB Native Ads 1.2 technical details
- Documented compact horizontal product card layout
- Added streaming response implementation details
- Updated architecture to show keyword/price extraction flow
- Clarified demo limitations vs production requirements
- Added integration guide for customers
- Followed documentation style from other use cases
v1.0 (2025-XX-XX): Initial generic documentation
- Original MCP-focused documentation

Conclusion

IAB Native Ads enable chat publishers to monetize conversational product discovery through standard, embedded product listing ads. This approach ensures:

Industry Standard: OpenRTB 2.6 + IAB Native Ads 1.2
Easy Integration: Works with all major ad platforms
Conversational: Natural language product queries with price filtering
Embedded Experience: Products naturally integrated within chat flow
Streaming Responses: Word-by-word text display for responsive UX
Strong Revenue: Performance-based monetization with high-intent queries
Price-Aware: Natural language price constraint extraction
Responsive Design: Adapts to desktop and mobile

By using standard IAB Native Ads 1.2 with natural language price filtering, you can generate revenue from conversational product discovery while maintaining a high-quality user experience and broad compatibility with the advertising ecosystem.

Live Demo​

Executive Summary​

Use Case Example​

Why IAB Native Ads for Chat Products?​

1. Industry Standard​

2. Structured Product Data​

3. Price Filter Extension​

4. LLM-Driven Intent Extraction​

5. Privacy-First Design​

6. Rich Analytics​

7. Quality Control​

Architecture Overview​

Natural Language Price Filtering​

Supported Price Patterns​

Price Filter Extension​

IAB Native Ads Asset Mapping​

Product Listing Native Assets​

Native Request Structure​

OpenRTB Integration​

Bid Request Structure​

Determining Number of Product Cards​

Bid Response Structure​

Product Display Integration​

Important: Products Are Embedded in Chat Messages​

Parsing Native Ad Response​

Rendering Product Cards​

Tracking Implementation​

Integration Workflow​

1. LLM Extracts User Intent​

2. Build OpenRTB Bid Request​

3. Call Affinity AI Bid API​

4. API Returns Product Ads​

5. Render Products in Chat​

6. Fire Tracking Pixels​

Benefits for Chat Publishers​

Revenue Generation​

User Experience​

Technical Advantages​

Analytics & Optimization​

Implementation Considerations​

Quality Control​

Privacy & Compliance​

Performance​

Native Ad Parsing​

API Response Handling​

No-Bid Scenarios​

Partial Bids​

Error Responses​

Related Use Cases​

Related Documentation​

OpenRTB Components​

Context Enhancement​

External Standards​

Version History​

Conclusion​