Streaming inventory is sold against content — genres, franchises, moods — but described to buyers in audience terms. Cookieless Audience bridges the two at the level where it can be done honestly: the web footprint of content. Every genre, show and content brand has a constellation of domains — fan communities, recap and review sites, sports properties, podcast pages — whose audiences are pre-computed in a 102M-domain database using fixed vocabularies aligned with IAB Audience Taxonomy 1.1. That gives streaming platforms a documented, code-level answer to “who is the audience for this content?” — for inventory packaging, ad-sales narratives and second-screen planning. This is domain- and content-level intelligence, not ACR or device-graph measurement, and we are explicit about that scope.
CTV ad platforms package inventory by content: a true-crime slate, a live-sports window, a comfort-food cooking block. Buyers, meanwhile, plan in audience language — demographics, interests, purchase intent. Connecting the two usually means leaning on panels that cover only the biggest titles, or on device-level identity graphs with their own coverage and privacy constraints.
There is a third source of evidence, systematically underused: every piece of content has a web footprint. A documentary genre has recap sites, subreddit-adjacent fan wikis, review blogs and podcast pages. A sports property has team sites, fan forums and stats communities. The people who read those domains are, to a useful approximation, the engaged audience of that content — and this database has already profiled all of them, with coded demographics, 285 sub-interests including twelve INT.television.* and thirteen INT.movies.* genre codes, 283 purchase-intent segments and resolved personas.
Mapping content to its web footprint and reading the aggregate audience profile gives platform teams a defensible, checkable audience narrative for any genre or franchise — including niche content that no panel will ever cover.
INT.television.factual_tv, demographics, intentDomain-level in, content-level out. No device IDs or viewing logs are involved at any step.
The same mapping serves the ad-sales, inventory and audience-insight sides of a streaming platform.
Group slates and genres into sellable packages defined by coded audience attributes — “the family_young_children co-viewing package”, “the high-income factual package” — with the domain-level evidence attached. See CTV content-audience mapping.
Replace “our viewers are engaged and upscale” with vocabulary values a buyer’s analyst can check against the published taxonomy — including which PI.* purchase-intent segments concentrate around each genre’s footprint, which is precisely the advertiser-category match sales teams need.
The web domains in a content footprint are themselves plannable inventory. Platforms and their agency partners use the same list to extend a CTV campaign to the open web where the genre’s audience already reads — the workflow described in media planning by persona.
The unit of work is a content footprint: a set of domains associated with a genre, franchise or slate. The real-time API adds per-URL granularity when a single large domain hosts many content verticals.
List the domains around the content: fan sites, review and recap properties, genre communities, sports or franchise sites.
Look up each domain in the database and collect its coded demographics, interests, intent segments and persona.
Roll the domain profiles up to the slate. Keep a confidence floor (medium+) so the narrative rests on well-evidenced attributes.
Read the over-indexing PI.* segments as an advertiser-category shortlist for the package’s sales sheet.
Attach the coded narrative to the inventory package; re-run the identical aggregation at each quarterly refresh.
A streaming platform wants an audience narrative for its true-crime package. The team assembles the genre’s web footprint — case-file blogs, fan wikis, recap sites, podcast companion pages — and aggregates their domain profiles. The concentrated attributes:
| Field | Concentrated value (code) | Reads as | Confidence |
|---|---|---|---|
| interest | INT.television.factual_tv | Factual / documentary TV | high |
| interest | INT.movies.crime_and_mystery_movies | Crime & mystery | high |
| interest | INT.news_politics.crime | Crime news | medium |
| age_bracket | 25_34, 35_44 | 25–44 core | high |
| gender_skew | female_lean | Skews female | medium |
| purchase intent | PI.arts_entertainment.music_and_video_streaming_services | In-market for streaming services | high |
| purchase intent | PI.arts_entertainment.radio_and_podcasts | Podcast listeners | medium |
These are complementary instruments. We state plainly what this dataset is and is not.
| ACR / device-graph data | Panel measurement | Domain-level content mapping (this dataset) | |
|---|---|---|---|
| Unit of observation | Device / household viewing | Recruited viewer sample | Web domains in a content footprint |
| Answers best | Who watched what, where | Reach & ratings currency | Who the engaged audience of a genre is, in coded attributes |
| Niche / long-tail content | Depends on device coverage | Thin below major titles | Strong — niche fan domains are profiled like any other |
| Advertiser matching | Behavioural, identity-bound | Demographic, coarse | 283 purchase-intent segments as category shortlists |
| Privacy surface | Device identifiers, consent-managed | Panel consent | No PII, no device IDs — domain-level only |
| Not suitable for | Open-web planning | Niche audience depth | Impression-level or per-stream measurement — by design |
No, and we don’t claim it does. It describes the audiences of the web domains associated with content — fan communities, review sites, genre properties. That is a distinct, complementary signal: the engaged-audience profile of a genre or franchise, useful precisely where viewing data is thin (niche content, pre-launch slates, competitive titles you have no logs for). Platforms typically use it alongside their first-party viewing data, not instead of it.
No. The dataset is domain-level and the real-time API is page-level, both built for planning and analysis. Neither classifies impressions in the bidstream, and CTV impressions have no URL to classify in the first place. The activation path is upstream: audience-defined inventory packages, sales narratives, and open-web second-screen domain lists.
As granular as the web footprint you define. Genre-level maps (true crime, reality, live sports) use dozens to hundreds of domains and are the most robust. Franchise-level maps work when a show has a substantive fan-site ecosystem. For large multi-vertical domains, the real-time API classifies individual URLs so a single fan hub’s sections can be separated. Below that — individual episodes, moods — the web evidence gets thin and we’d advise against over-claiming.
Content footprints skew long-tail, so most platform teams take the top-1M file ($1,990 one-time, $590/quarter, instant checkout) rather than the top-100k ($490). Entertainment-vertical slices run $190–$490. Teams that want the full 102M corpus, custom genre rollups or a recurring feed contact us for a quote — custom licensing starts at $15,000/year.
The full workflow guide with more worked examples.
Turn footprint domains into second-screen plans.
The same mapping logic applied to out-of-home screens.
How content owners document their own audiences.
Pick a genre, look up its fan domains in the demo dashboard, and see the audience narrative assemble itself — then license the tier that covers your catalogue.
Open the audience demo See database pricing