Topsort Prompts is an ad format that helps marketplaces monetize chatbot conversations by adding sponsored product tiles to chatbot responses. Marketplace chatbots are now a major product discovery surface. Shoppers use them for high-intent questions like gift ideas, replacements, and product comparisons. Topsort Prompts turns those moments into auction-eligible inventory while keeping recommendations useful and relevant.Documentation Index
Fetch the complete documentation index at: https://docs.topsort.com/llms.txt
Use this file to discover all available pages before exploring further.
Why It Matters
- Monetizes chatbot discovery
- Preserves shopper trust with blended sponsored and organic recommendations
- Captures intent semantically, not only through strict keyword overlap
- Uses existing sponsored listings setup and operations
What Topsort Prompts Does
When a shopper asks a product question in a chatbot connected to Topsort MCP, Topsort evaluates the prompt against active campaigns using semantic matching. When a marketplace connects its chatbot to Topsort’s Topsort Prompt server, Topsort returns products when users query for them. If the user query matches eligible sponsored products, sponsored tiles are returned first. Otherwise, Topsort returns organic recommendations. This creates value for all sides:- Shoppers get useful recommendations
- Advertisers reach high-intent users
- Marketplaces monetize conversational discovery
How It Works
- Marketplace enables Conversational Prompts in Topsort UI
- Marketplace creates Topsort Prompt campaigns with prompts, products, targeting, and timing
- Chatbot sends shopper prompts through Topsort MCP
- Topsort semantically matches prompt intent to active campaigns
- If a query matches eligible sponsored products, Topsort returns sponsored tiles first and then organic tiles; otherwise, it returns organic recommendations
Semantic Matching Example
A campaign targetingsports watch can match prompts like:
- “Help me find a watch for running”
- “I need a fitness watch under 60”
- “Can you recommend a workout watch?”