Cited Inventory API — the domains AI engines cite, and who sells them

​

What it is for

• Media companies and sales houses: see which of your represented titles are AI-cited, and which high-authority domains in your market you do not yet represent.
• Agencies and DSPs: build curated deal lists or contextual targeting from domains whose authority is validated by the answer engines themselves. Pages that AI engines cite receive the clicks from AI answers — human traffic in active research mode.
• Advertisers: pull your own tracked brands’ cited sources into your planning tools, with the seller of each source identified.

​

Authentication and key scopes

curl -H "Authorization: Bearer bsk_partner_..." \
  "https://app.trybluemoon.com/api/partner/v1/cited-inventory?market=CH"

​

Endpoint

GET /api/partner/v1/cited-inventory

​

Response

{
  "scope": "organization",
  "organization": { "id": "…", "name": "Acme AG" },
  "generatedAt": "2026-08-18T11:40:00.000Z",
  "windowDays": 90,
  "filters": { "market": "CH", "minCitations": 10, "sellableBy": null },
  "count": 142,
  "hosts": [
    {
      "host": "example-review-site.ch",
      "citations90d": 49,
      "trendPct": 12.5,
      "markets": { "CH": 41, "GLOBAL": 8 },
      "organizations": 4,
      "platforms": ["chatgpt", "gemini", "perplexity"],
      "sellableBy": [
        {
          "house": "example-sales-house",
          "confidence": "confirmed",
          "relationship": "DIRECT",
          "evidence": ["google.com, pub-0000000000000000, DIRECT, f08c47fec0942fa0"]
        }
      ],
      "adSystems": [
        { "adSystem": "google.com", "relationship": "DIRECT", "accounts": 2 },
        { "adSystem": "rubiconproject.com", "relationship": "RESELLER", "accounts": 1 }
      ]
    }
  ]
}

• citations90d — how many times AI answers cited this domain in the last 90 days.
• trendPct — last 30 days vs the 30 days before, in percent; null when the earlier window is too small for a meaningful trend.
• markets — citations broken down by the country the answers were collected for. Bluemoon collects answers with real geo targeting per market, not prompt wording.
• sellableBy — sales houses from the directory below whose footprint appears in the domain’s public ads.txt. confidence is confirmed for account-level evidence and probable for a comment-only mention; when several houses match, they are ordered strongest first (confirmed DIRECT before resellers). Every claim carries the raw ads.txt lines as evidence — the source is public and auditable.
• adSystems — the domain’s raw monetization profile (which SSPs, DIRECT or RESELLER, how many seller accounts). If your sales house is not in the directory yet, match your own account ids against this field.

​

The sales-house directory

​

Using it with OpenRTB

• Curated deals: package the filtered host list as a private marketplace deal in your SSP; the deal travels as imp.pmp.deals[].id in bid requests. Refresh from the feed on your own schedule; your SSP owns the deal.
• Contextual segments: attach page-level segments (site.content.data[].segment[]) such as AI-cited source via your SSP or a Prebid RTD module. The data describes pages, never users — there is no personal identifier anywhere in this product, by construction.

​

Freshness, privacy, and limits

• The global feed is a pre-computed snapshot, refreshed weekly; generatedAt tells you its age. Organization-scoped responses are computed from your live data behind a short cache. Either way, polling is always cheap and always fast.
• The global feed is domain-level only, and a domain is published only when it clears a k-anonymity gate (cited across at least 2 independent organizations, or 50+ citations) — it can never reveal what any single Bluemoon customer tracks. Organization-scoped keys see only that organization’s own data.
• No user data, no audience segments, no identifiers. Domain-level facts derived from public AI answers and public ads.txt files.