For advertisers
Reach people while they are describing what they need
Search gives you a keyword. Social gives you a profile. An AI conversation gives you a paragraph in which someone states their situation, their constraints and their budget — unprompted, in their own words, at the moment they are deciding.
Monetzly places sponsored recommendations inside the responses of AI assistants that have opted into advertising. This page covers what the inventory is, how matching works, and the limits we hold on both sides.
What the inventory is
The unit is a sponsored mention inside an assistant's answer: a specific, labelled product recommendation that belongs in the response the user was already getting. It is not a banner beside a chat window and it is not an interstitial.
- Placement lives in the response text, in the assistant's own voice, always labelled as sponsored.
- Publisher apps are LLM-native products — assistants, copilots and chatbots — not general web pages.
- Frequency is capped per session, so a user sees a placement rarely rather than on every turn.
- A placement only runs when the conversation expresses a need the campaign can legitimately serve.
How matching works
Matching is contextual. It reads the conversation happening right now — not a profile assembled from a user's history across sites.
- You describe the need your product answers, in plain language, plus the categories and exclusions you want.
- During a conversation, the ad layer evaluates whether the user's expressed intent genuinely matches that description.
- On a match above threshold, the sponsored mention is injected into the response stream with a disclosure label.
- Impression, click and session data are attributed and reported back per campaign.
What this means practically
Intent is expressed rather than inferred. A user writing "I need a pan that won't stick for a gas hob, under $60" has handed over the brief. There is no cross-site tracking involved, and no behavioural profile to buy against — the conversation is the targeting signal.
Brand safety runs in both directions
Publishers control which advertisers can appear in their app; advertisers control which conversations they will appear in. Both sets of controls are enforced before a match, not audited afterwards.
| Control | Who sets it | Default |
|---|---|---|
| Category exclusions | Advertiser and publisher | Publisher's blocked list always wins |
| Sensitive contexts (health, finance, legal, minors) | Platform policy | No placements unless explicitly enabled |
| Emotional context | Platform policy | Suppressed — distress is not an audience |
| Frequency per session | Platform policy + publisher | Capped low by default |
We also refuse placements that would require the assistant to give advice it would not otherwise give. If the honest answer to "do I need this?" is no, no campaign changes that. It is a limit on short-term revenue and the reason the channel keeps working.
Measurement
- Impressions and clicks attributed per session, so a conversation is the unit of analysis rather than a page view.
- Match context reporting: which kinds of expressed intent your campaign served, so you can tighten the brief.
- Standard destination-side tracking still applies — clicks land on your URLs with your own parameters intact.
Who this fits — and who it does not
Good fit
- Products people research before buying, where a specific recommendation is genuinely useful.
- Categories with a clear, describable need: tools, gear, software, travel, services.
- Advertisers comfortable with contextual targeting and low, high-quality volume over reach.
Poor fit
- Pure brand-awareness campaigns measured on impression volume — the format is deliberately low-frequency.
- Categories that depend on behavioural profiles or retargeting.
- Regulated categories in sensitive contexts, which we suppress by default.
Read the rules we hold ourselves to
Talk to us about a campaign
Tell us the need your product answers and the categories you want to stay out of. We will tell you honestly whether the inventory exists yet.