AI answer ad tracking

Answer engines have started selling placements inside the answer itself — sponsored slots in ChatGPT, sponsored follow-ups in Perplexity, paid units in Google's AI surfaces, ads in Copilot. Ads, inside AI Visibility, reports what's observable about that: for AI answers already collected on your tracked prompts, it shows whether a paid placement appeared, who bought it, and where it sat relative to your own organic citation.

This is observation only. SearchChamp doesn't buy an ad, bid, or retarget on your behalf — this is the visibility layer that tells you it's happening, not a way to run or manage AI answer ads.

It costs nothing extra to run. Every answer this feature analyses was already fetched and stored by your AI Visibility tracker for its own citation tracking — nothing is probed a second time, and nothing counts against a separate quota.

What we can see, and why it differs by engine

Not every engine tells us the same thing, and that's a fact about what each engine's API actually returns — not a gap SearchChamp expects to close evenly over time. There are three tiers:

• Structured detection (Google's AI surfaces, Copilot) — the vendor data already received for these engines carries explicitly labelled paid items, read directly. Highest-confidence tier: no guessing, no model involved.
• Classifier-based detection (ChatGPT, Perplexity) — these engines sell placements in their consumer products, but their API answers are prose with no structured ad fields. A pass over the stored answer text looks for disclosure language. Anything not confident is labelled "possibly sponsored" rather than counted as a confirmed ad.
• No coverage (Claude, Gemini, Grok) — these engines' APIs expose nothing about paid placements. The panel says so explicitly: "this engine does not expose ads."

That third line matters, so it's worth stating plainly: "this engine does not expose ads" is not the same as "no ads ran there." It means there's no way to look. SearchChamp says that outright rather than show a "0 ads" number for an engine it can't see into — a zero there would look like a finding, and it isn't one. The only zero this feature ever shows as a finding is a "no ads detected" result from an answer that was successfully analysed and genuinely had no paid placement in it.

What "possibly sponsored" means

For ChatGPT and Perplexity, the classification pass reads a stored answer and looks for a paid placement. Nothing is labelled an ad unless the answer itself gives a reason to — a disclosure word like "Sponsored," "Ad," "Advertisement," "Paid partnership," or "Promoted," near a named advertiser. The classifier can only make that signal weaker, never stronger: it can talk itself out of a weak, ambiguous disclosure, but it can never manufacture an ad label out of an answer with no disclosure text at all.

When the evidence is strong enough, the placement is labelled sponsored. When it's present but not strong enough to assert, it's labelled possibly sponsored — you'll see that literal wording in the panel and the advertiser table, kept in its own group, and it never contributes to an "ads appeared" or "who's advertising against you" count.

Long answers can hide a disclosure

The stored answer text behind this feature is capped in size — very long answers are truncated before a disclosure check happens. If the sponsorship notice sat past that cut point, it won't be seen, and the answer is flagged as partially analysed rather than silently treated as complete.

The near-miss list: when a competitor sponsors above you

If you're cited organically on a tracked prompt and a different advertiser holds a paid placement above your citation, that prompt shows up in a dedicated near-miss list — the clearest sign a competitor is buying attention on a query you already earned organically. Two things keep this list honest:

• If the paid placement's advertiser domain is your own, it's excluded — you can't out-rank yourself.
• If the only evidence is "possibly sponsored," the prompt is excluded from this list. A maybe never turns into "a competitor bought a slot" here.

What this is not

• Not ad buying, bidding, or budget management of any kind.
• Not a bid-intelligence or estimated-CPC tool.
• Not a retargeting integration.
• Not a promise that an engine showing "does not expose ads" today will stay that way — if an engine starts exposing ad data, coverage for it can be added, but nothing is inferred about it in the meantime.

How to get notified

Ad tracking will live at AI Visibility → Ads once it's enabled, on a Business plan or above. To be told when it's live, watch the What's New page — there's nothing to set up in the meantime, since it rides on answers your tracker already collects.

FAQs

Can I use this today? No — it isn't enabled yet. This page documents what it will show so you know what's coming; there's no tab to find and no setup to do right now.

Will this cost extra AI Visibility credits? No, once it launches. Everything here is designed to ride on answers your AI Visibility tracker already fetched for citation tracking — no engine gets queried an extra time, and it won't consume any additional tracking allowance.

Why don't Claude, Gemini, and Grok show any ad data? Their APIs don't expose anything about paid placements, so there's nothing to detect. The panel says this explicitly rather than showing a misleading zero.

Can I trust a "possibly sponsored" label the same as "sponsored"? Treat it as a signal, not a confirmed fact — it's kept in its own group in the UI and deliberately excluded from confirmed-ad counts and the near-miss list.

Related articles

• AI Visibility reference
• Brand integrity fact sheet
• Agent Experience (AXP)