Glossary
What is prompt inventory?
Prompt inventory is the set of user prompts on an AI surface that are eligible for monetisation, broken down by topic, intent strength, commercial relevance and exclusion rules — the AI-era analogue to page inventory.
How it differs from page inventory
A publisher's page inventory is knowable in advance. The pages exist, their number is roughly predictable, and each can carry several placements.
Prompt inventory is created by users in real time and is unknowable until it arrives. It cannot be increased by producing more of it — the publisher does not write the prompts. And most of it is not sellable at all, because most conversations have no commercial dimension or are ones where advertising would be inappropriate.
The practical consequence is that inventory forecasting is a demand-side estimation problem rather than a supply-side inventory count, and the numbers are far smaller than page-based intuitions suggest.
How a publisher segments it
- By topic, which is what advertisers actually buy against.
- By [intent density](/resources/glossary/intent-density), which is what sets the price.
- By decision stage — exploring, comparing, deciding — because the same topic behaves very differently at each.
- By exclusion. The categories where no advertising is permitted regardless of what it would pay: health and legal concerns, distress, minors, anything sensitive.
- By format eligibility, since not every eligible prompt suits every placement type.
Exclusion is the first decision
On a web page the default is that a slot is sellable and exceptions are carved out. On an AI surface the workable default is the opposite: a conversation is not monetisable unless it clearly qualifies.
That inversion is what protects the product. A user who encounters advertising while asking about a medical symptom or a financial difficulty has learned something about the assistant that no amount of relevance elsewhere repairs. Publishers who have made this work run conservative eligibility rules and treat them as a product constraint rather than a revenue ceiling to be optimised against.
Measuring it
The natural unit is per query rather than per thousand impressions, because queries are what the publisher produces and only a fraction are monetisable. Revenue per query is the metric most operators track, with eligible-prompt rate — the share of conversations that qualify at all — as the supply-side number underneath it.
Common questions
Can a publisher increase prompt inventory?
Only by growing usage or by loosening eligibility. The second raises short-term revenue and is the most reliable way to damage the product, so most serious operators treat eligibility as fixed and grow the base instead.
What proportion of prompts are monetisable?
It varies enormously by surface. A shopping assistant will be far higher than a general one. Publishers running disciplined eligibility rules typically find it is a small minority of total conversations.
More in ai and conversational advertising
How buying, selling and measurement work when the surface is a conversation.
- AdCP (Advertising Context Protocol)
- Agent-to-Agent Attribution (A2A attribution)
- Agentic advertising
- Agentic conversion
- AI ad network
- AI DSP (AI demand-side platform)
- AI media
- AI open web
- AI publisher
- AI share of voice (AI SOV)
- AI SSP (AI supply-side platform)
- AI walled gardens
- Answer insertion
- Artificial intelligence (AI)
- Branded agent
- ChatGPT Ads
- Citation bidding
- Conversational ads
- Conversational frequency cap
- GEO / AEO (Generative engine optimization / Answer engine optimization)
- Hallucination liability
- Intent density
- Machine learning
- MCP (Model Context Protocol)
- Post-click ad economy
- Sponsored follow-up
- Sponsored intelligence (SI)
- Sponsored prompt
- Synthetic audience
- Trust graph