Cookies remain on Chrome, but Safari, Firefox and iOS block them — so on roughly 40%+ of the inventory a DSP can buy, identifier-based audience targeting simply does not fire. Cookieless Audience gives buy-side teams a planning layer for exactly that supply: pre-computed audience attributes for 102 million domains, aligned with IAB Audience Taxonomy 1.1, plus a real-time API for page-level audience segmentation. Build domain lists by audience, research supply before a campaign, and enrich delivery logs afterwards. To be precise about the boundary: this is planning and analysis data — not impression-level pre-bid classification in the bidstream.
On Chrome, a DSP can still lean on third-party cookies. On Safari, Firefox and iOS it cannot — and that share of supply is often exactly where the audience lives, especially for affluent, mobile-heavy and Apple-skewed segments. Without identifiers, campaigns on that inventory degrade to run-of-exchange plus contextual guardrails.
Domain-level audience data restores an audience view at the level a buyer can actually control: the supply itself. Every domain in the corpus carries coded demographics (8 age brackets, 6 income bands, 14 life stages and more), 285 sub-interests, 283 purchase-intent segments, B2B firmographics and persona assignments — each with banded confidence. Filter the corpus by your brief and you get a defensible, inspectable domain list where the target audience is known to concentrate.
That list becomes an inclusion list, a PMP negotiation agenda, or the audience evidence in a plan — and it works identically on cookied and cookieless traffic, because it never referenced an identifier in the first place.
// Brief-to-list query, conceptually // "EV launch: 35-54, upper income, auto-intenders" SELECT domain FROM audience_db WHERE purchase_intent CONTAINS "PI.auto_ownership.new_vehicles" AND interests CONTAINS "INT.automotive.auto_technology" AND age_bracket IN ("35_44", "45_54") AND income_level IN ("upper_middle", "high") AND confidence = "high"; // Output: a domain list for inclusion lists and // deal negotiation — built before the campaign, // not decided per impression in the bidstream.
The dataset supports the work that happens around the bidder — before campaigns launch, while they run, and after they finish.
Translate a brief into attribute filters and export the qualifying domains as an inclusion list or deal-negotiation target. It is the cookieless counterpart of an audience segment: instead of finding users, you find the properties their attention concentrates on. See media planning by persona.
Profile unfamiliar publishers before committing budget: who reads this property, does its audience match the brief, how does it compare with the sites already on the plan? The per-URL API extends the same check to individual sections and landing pages.
Join delivery logs against the domain file to explain where a campaign actually ran in audience terms — which personas, interests and income bands the delivered domains represent. The workflow is documented under ad log enrichment.
The data does its work before and after the auction. The auction itself stays yours.
Map the target audience onto coded attributes — personas, INT.*, PI.*, demographics, firmographics. The taxonomy page lists every field.
Query the licensed file with your attribute logic, gated on medium or high confidence.
Load the result as inclusion lists in your seats, or hand it to supply partners as the basis for PMPs and curated deals.
Mid-flight, profile unfamiliar URLs surfacing in logs with the per-URL API — same vocabularies, page-level resolution.
Post-campaign, join delivered domains back to audience attributes for reporting, learning and the next plan.
The database describes domains; the real-time API describes pages. Both are for planning, research, curation and enrichment. Neither sits in your bid path, and we never claim impression-level pre-bid capability. If a vendor decision needs a millisecond-latency bidstream classifier, that is a different product category — this dataset is what you plan and evaluate with, and it deliberately stays out of the auction.
An automotive client is launching an electric model. The audience brief: 35–54, upper income, technology-curious, actively considering a new vehicle — with heavy expected consumption on iOS and Safari, where identifier targeting will not fire. The team builds the plan from coded attributes.
| Brief element | Coded filter | Reads as |
|---|---|---|
| In-market for a new car | PI.auto_ownership.new_vehicles | Domains whose audience shows new-vehicle purchase intent |
| Technology-curious | INT.automotive.auto_technology | Auto-technology interest — the EV angle |
| 35–54 | age_bracket ∈ {35_44, 45_54} | Readership concentrated in the target brackets |
| Upper income | income_level ∈ {upper_middle, high} | Income bands matching the price point |
| Evidence gate | confidence = high on intent | Only strongly supported intent classifications survive |
| Identifier-based audience targeting | Contextual pre-bid segments | Audience-informed domain lists (this dataset) | |
|---|---|---|---|
| Decision point | Per impression, in the auction | Per impression, in the auction | Before the campaign — list and deal construction |
| Cookieless coverage | Fails where IDs are absent (Safari, Firefox, iOS) | Full | Full — keyed on the domain, not the user |
| What it knows | The individual user (where resolvable) | The page’s content | The property’s audience: demographics, interests, intent, personas |
| Granularity | Impression | Impression | Domain in the dataset; URL via the real-time API for research |
| Best used for | Retargeting and 1:1 where IDs persist | Suitability and adjacency at bid time | Planning, allowlists, PMP targets, log analysis — and it composes with both others |
No, and we state that deliberately. The database is domain-level and the real-time API is per-URL for planning and analysis; neither is built for millisecond bid-path classification, and we never claim impression-level pre-bid capability. What the data produces — audience-informed inclusion lists and deal targets — is then applied by your bidder through completely standard mechanisms.
A contextual segment describes what a page is about; this dataset describes who reads a property — age brackets, income bands, life stages, 285 sub-interests, 283 purchase-intent segments, personas. A cycling blog and a cycling retailer’s buying guide are contextually similar but can have different audiences and very different intent profiles. The two layers are complementary, and many teams run both.
The domain file refreshes quarterly on paid refresh plans, which fits the cadence of planning and list-building. For anything that moves faster — a new publisher appearing in logs, a specific landing page, a section of a large site — the real-time API classifies individual URLs on demand using the same fixed v1.0 vocabularies, so spot checks stay consistent with the plan.
From the buy side, the same taxonomy alignment works in reverse: when SSPs and curation platforms declare seller-defined audiences under IAB Audience Taxonomy 1.1 or offer audience-curated Deal IDs, your own copy of the data lets you evaluate those claims — compare a deal’s domain list against the attributes you expect, before and after committing spend. It makes the buy side an informed counterparty rather than a price-taker.
Brief-to-domain-list planning in full detail.
Reaching the traffic where identifiers never fire.
Explaining delivery in audience terms after the flight.
How the sell side packages the same data.
Profile any domain in the audience demo, or license the file and turn your next brief into an audience-informed inclusion list.
Open the audience demo See database pricing