Conversaic

Labeling and guardrails

Guardrails define when a placement is allowed to render. Configure them before launch, then review them whenever the placement surface expands.

Disclosure labels

Label Use when Visibility rule
Sponsored The recommendation is commercially funded. Show on the card before click.
Affiliate recommendation A downstream action may generate affiliate revenue. Show near the title or CTA.
Partner recommendation The publisher prefers partner-language disclosure. Keep the commercial relationship clear.

Suppression rules

Weak match
Return no placement when the request does not contain a useful recommendation moment.
Sensitive context
Suppress monetization for health, legal, financial distress, personal crisis, or other restricted topics.
Blocked category
Publisher policy can block an otherwise eligible commercial offer.
Tracking unavailable
Do not launch production placements when impression or click reporting is not working.

Policy configuration

Policy Example
Allowed categories developer_tools, productivity, education
Blocked categories medical, gambling, crypto, adult
Allowed placements answer_footer_sponsor, comparison_card
Blocked contexts Sensitive prompts, support escalation, user safety flows
Rollout scope Dev only, one placement, or one publisher traffic segment

Review checklist

  1. The label is visible before click.
  2. The recommendation matches the user request.
  3. Publisher policy allows the category and surface.
  4. Sensitive contexts return no placement.
  5. The event trail includes impression, click, offer, placement, and publisher identifiers.
Audience: Developer-tool and conversational AI app publishers Updated: 2026-04-24