A curation business lives or dies on one question: for any Deal ID you sell, can you say — credibly, at scale, across every SSP you touch — who the audience behind those domains is? Cookieless Audience answers it as infrastructure: pre-computed audience segmentation for 102 million domains, every attribute drawn from fixed vocabularies aligned with IAB Audience Taxonomy 1.1, plus a real-time API for page-level segmentation of individual URLs. It is the substrate under a curated marketplace — the layer that turns supply footprints into audience-defined deals, on cookied and cookieless traffic alike.
Building one good curated deal is a research project: pick a theme, find the domains, justify the audience claim, keep the list current. Building a marketplace of hundreds of deals across tens of thousands of domains — and defending every audience claim to a buyer — is a data problem no ad-ops team can solve by hand.
The Cookieless Audience database makes the knowledge side of curation a query instead of a project. Every domain carries coded demographics (8 age brackets, 6 income bands, 14 life stages, household, employment, urbanicity), 29 interest groups with 285 sub-interests, 34 intent groups with 283 purchase-intent segments, B2B firmographics and one of 1,667 deterministic personas — each attribute with banded confidence. Deal construction becomes boolean logic over those fields, repeatable across every SSP integration you operate.
And because there is no PII and no identifier anywhere in the pipeline, the deals you build work identically on the 40%+ of traffic where Safari, Firefox and iOS block third-party cookies — cookies remain on Chrome, but your product no longer depends on them anywhere.
// A deal definition, as segment logic // Theme: "Affluent frequent travelers" { "deal_theme": "affluent_frequent_travelers", "rules": { "purchase_intent": [ "PI.travel.hotels_and_resorts", "PI.travel.air_travel" ], "income_level": ["high", "affluent"], "confidence_min": "medium" } } // Run against the licensed file -> domain set. // Attach to Deal IDs in each SSP seat you curate.
Curation platforms use the data at catalog level, deal level and buyer level — the same coded attributes flow through all three.
Join the domain file to the total supply footprint reachable through your SSP and exchange integrations. Every reachable domain gets a consistent audience profile — including the long tail where no publisher declaration or first-party data will ever exist.
Express each marketplace product as attribute logic — personas, INT.*, PI.*, demographics, firmographics — and regenerate domain sets programmatically. A hundred deals become a hundred queries, not a hundred research projects. Mechanics: inventory curation.
Every deal ships with its rule set and confidence bands — a one-sheet a buyer can interrogate. Crosswalk attributes to IAB Audience Taxonomy 1.1 nodes and the same logic backs seller-defined audience declarations on the curated supply.
The substrate is a file you own and query — every downstream artifact regenerates from it.
Top 100k or top 1M domains self-serve, vertical and country slices, or 5M up to the full 102M corpus for marketplace-scale operations — see pricing.
Intersect with the domains your SSP and exchange integrations can actually transact, so every deal is buildable on day one.
Write each product as boolean rules over coded attributes, gated on confidence bands. Version the rules — the vocabularies are fixed at v1.0, so logic stays stable.
Materialize domain sets per deal and push them to Deal IDs across your seats. Regeneration is a re-run, not a rebuild.
For large multi-topic publishers, use the per-URL API to curate at section level — the finance vertical of a news site can sit in a different deal than its sports vertical.
A curation platform wants a travel-audience product for luxury-hospitality and airline buyers, spanning every SSP it curates on. The deal is defined once, as attribute logic, and materialized everywhere.
| Rule (coded values) | Reads as | Why it’s in the logic |
|---|---|---|
PI.travel.hotels_and_resorts or PI.travel.air_travel | In-market for hotels, resorts or flights | The commercial core — intent is what the buyer pays a premium for |
INT.travel.adventure_travel (optional boost) | Adventure-travel interest | Sub-theme for an experiential-travel variant of the deal |
income_level ∈ {high, affluent} | High-income readership | Matches the luxury price point of the advertisers |
audience_type = b2c | Consumer-facing properties | Excludes trade and B2B travel media from a consumer deal |
confidence ≥ medium, high for the flagship tier | Evidence gate | Two deal tiers: broad reach at medium, flagship at high confidence |
segtax: 4 declarations. Next quarter’s refresh re-scores the corpus; re-running the same logic updates every deal in the marketplace at once.| Build your own classifier | Publisher-declared / first-party data | Licensed domain-level dataset (this one) | |
|---|---|---|---|
| Time to first deal | Months of crawling, modeling, QA | Per-publisher negotiation | Days — the file arrives pre-computed |
| Coverage | Whatever you crawl and maintain | Only cooperating publishers; long tail absent | 102M domains, uniform schema, long tail included |
| Taxonomy alignment | Yours to design and defend | Varies per publisher | Built aligned with IAB Audience Taxonomy 1.1, fixed v1.0 vocabularies |
| Freshness | Your pipeline, your ops burden | Publisher-dependent | Quarterly refresh; per-URL API between refreshes |
| Cost shape | Engineering headcount, ongoing | Revenue shares, integrations | One-time license + quarterly refresh; instant-buy tiers from the pricing page |
Most desks start with the top 1M domains — available as an instant card purchase with immediate download — because it covers the supply that transacts meaningfully in most markets. Platforms curating deep long-tail or international supply license 5M up to the full 102M corpus under a custom agreement. Vertical and country slices exist for specialist marketplaces. Details are on the pricing page.
Yes. The domain file is the substrate for property-level deals; the real-time API applies the same v1.0 vocabularies to individual URLs, so you can curate sections of large publishers separately — putting a news site’s personal-finance vertical in an investing deal without dragging in its celebrity coverage. Domain-level and URL-level outputs share one schema, so mixed-granularity deals stay coherent.
Every attribute is a coded value from a fixed, published vocabulary with a banded confidence score, so a deal one-sheet can state its exact inclusion logic — not a vague audience description. Buyers holding their own copy of the data can re-run the logic and check the domain list themselves. That inspectability is a sales asset: curated products survive procurement when the evidence is reproducible.
No. The dataset is planning and construction infrastructure: it determines which domains belong in a deal and documents why. Once a Deal ID exists, targeting and delivery run entirely in the SSPs and DSPs transacting it. We deliberately make no impression-level pre-bid claims — the product is the substrate, not the auction.
The deal-construction workflow step by step.
Declaring curated audiences under segtax 4.
Pricing supply by the audience it carries.
The platform-side view of the same substrate.
Test the attribute depth on any domain in the audience demo, then license the slice of the corpus your marketplace needs.
Open the audience demo See database pricing