Syndicate Links

Chatbot Publisher Programs

Chatbots are the most common form of AI agent already participating in commerce. Customer service bots, shopping assistants, onboarding agents, and developer documentation bots all make recommendations that influence purchasing decisions. A chatbot publisher program gives these bots the ability to earn commission on the conversions they drive.

Types of Chatbot Referrals

Customer Service Bots

Customer service chatbots handle product questions, troubleshoot issues, and suggest solutions — which often means recommending products, upgrades, or complementary services. When a support bot recommends a product and the customer buys it, that is an attributable referral.

Example flow:

Customer: "My headphones keep disconnecting from Bluetooth."
Bot: "That sounds like a Bluetooth 4.2 limitation. The Model X Pro uses Bluetooth 5.3
      and has 98% fewer disconnection reports. Would you like me to look into upgrading?"
Customer: "Sure, show me the options."
Bot: [tracks attribution event] → "Here are the upgrade options..."

The attribution event is submitted when the bot makes the specific product recommendation, not when the customer asks the initial question.

Shopping Assistants

AI shopping assistants — standalone products like Perplexity Shopping or embedded agents on e-commerce sites — exist specifically to help users find and purchase products. Every recommendation is a potential attribution event.

These agents typically operate at higher volume than customer service bots, processing dozens or hundreds of product recommendations per hour. The tracking infrastructure must handle this volume without adding latency to the conversation.

Developer Documentation Agents

Documentation agents (Kapa.ai, Mintlify, custom-built) answer developer questions and frequently recommend specific tools, libraries, or services. When a developer asks "What testing framework should I use?" and the agent recommends a specific paid tool, that is an attributable referral.

Developer tool recommendations have particularly high conversion rates because the agent is answering at the exact moment of decision — the developer is actively choosing a tool and the agent's recommendation carries significant weight.

Onboarding Agents

Product onboarding chatbots guide new users through setup and configuration. During this process, they often recommend complementary products, integrations, or paid upgrades. These recommendations are high-intent — the user is actively building their stack and receptive to suggestions.

Integration Patterns

Pattern 1: Recommendation-Time Tracking

The most common pattern. The chatbot submits an attribution event at the moment it makes a recommendation:

# Inside the chatbot's recommendation handler
def recommend_product(user_query, product):
    # Track the attribution event
    sl_client.track(
        merchant_id=product.merchant_id,
        product_id=product.id,
        metadata={
            "conversation_id": conversation.id,
            "query": user_query,
            "channel": "support_chat"
        }
    )

    # Return the recommendation to the user
    return f"I recommend {product.name}. {product.description}"

For chatbots that provide product links, the attribution token can be appended to the URL as a query parameter. This bridges the gap between the chat interface and the merchant's website:

https://merchant.com/product/123?sl_ref=slat_v1_...

The merchant's checkout reads the sl_ref parameter and submits it for verification at conversion time. This pattern works when the chatbot operates in a browser context where the user will click through to a web page.

Pattern 3: MCP Tool Integration

For chatbots built on MCP-compatible runtimes, the Syndicate Links MCP server provides attribution as a native tool. The chatbot calls track_referral as part of its tool chain:

Agent tool call: track_referral(merchant_id="linear", product_id="pro_plan")
Agent response: "I recommend Linear Pro for your team size. It handles..."

No custom API integration code required.

Commission Models for Chatbots

Different chatbot types align with different commission models:

Chatbot TypeTypical Commission ModelRate Range
Shopping assistantsPercentage of sale3-15%
Customer service botsFlat rate per upsell$5-50
Developer tool agentsFlat rate per signup$10-100
Onboarding agentsPercentage of upgrade5-20%

Fraud Prevention

Chatbot publisher programs introduce specific fraud vectors:

  • Phantom recommendations — bots claiming attribution for recommendations never shown to users
  • Recommendation stuffing — bots recommending products in every response regardless of relevance
  • Conversation manipulation — bots steering conversations toward high-commission products

Syndicate Links mitigates these through rate limiting per agent key, conversion matching (the merchant must independently confirm the purchase), and anomaly detection on recommendation patterns.