Select inventory with coded audience attributes and quality filters, then package it behind a Deal ID that means something. The 102M-domain database gives curation teams per-domain demographics, 285 interests, 283 intent segments, personas and confidence bands — so packages are reproducible queries, not hand-maintained spreadsheets.
Curation is ultimately list-building: deciding which domains belong in a package and defending that decision to a buyer. Most curation today relies on content categories and blocklists alone — the audience dimension lives in the sales deck but not in the selection logic.
INT.* interests + PI.* intent + quality filtersA defensible package combines an audience thesis with quality and suitability controls. All three layers come from the same per-domain record, documented on the audience segmentation taxonomy page.
Who the package reaches:
INT.* sub-interests and 283 PI.* intent segmentsWhat stays out:
medium+ or high only)audience_type filters (b2c / b2b / mixed)The filtered domain list becomes the inclusion list behind a Deal ID in your SSP or curation platform. The query itself is archived as the audit trail — fixed v1.0 vocabulary codes keep it meaningful quarter after quarter.
Six steps from package thesis to a live, refreshable Deal ID.
One sentence a buyer would pay for: “Premium reach against expecting and new parents with baby-product intent, brand-suitable, cookieless-inclusive.”
Restate the thesis as vocabulary codes — life stages, interests, intent segments — so the package is a query, not a vibe.
Set a confidence floor, restrict audience_type, limit to a corpus tier if the buyer wants head inventory, and layer IAB content categories for suitability.
Review the head of the resulting domain list, check the package is large enough to clear, and prune misfits — the list is readable, so this takes minutes.
Load the domain list as the inclusion list for a PMP or programmatic-guaranteed deal in your SSP or curation platform, and publish the one-sheet from the same field values.
Re-run the archived query against each database refresh. Domains that no longer qualify drop out; newly qualifying domains join.
A CPG buyer wants premium reach against expecting and new parents. Below is the package definition as written against the database — codes on the left, buyer-facing labels on the right.
| Layer | Filter (code) | Reads as |
|---|---|---|
| Audience | life_stage ∈ {expecting_parent, new_parent} | Expecting parents and new parents |
| Audience | household_composition = family_with_children | Family households |
| Interest | INT.family_relationships.parenting_babies_toddlers | Parenting Babies & Toddlers |
| Intent | PI.family_parenting.childcare, PI.cpg.edible | Childcare; edible CPG purchase intent |
| Quality | confidence ≥ medium | Well-evidenced attributes only |
| Quality | audience_type = b2c | Consumer-facing sites only |
| Suitability | IAB content category allow-list via the API | Family & parenting content, sensitive categories excluded |
Result: a domain inclusion list — parenting magazines, baby-gear review sites, pregnancy communities and family-lifestyle blogs — each row with full attributes and confidence bands.
How it ships: the list goes to the SSP as the Deal-ID inclusion list; the query goes to the buyer as the package definition. Preview the data in the audience demo.
| Run-of-exchange + blocklist | Content-category curation | Audience-attribute curation (this dataset) | |
|---|---|---|---|
| What defines the package | Everything not blocked | Site topic (e.g. “sports”) | Who reads the site: coded demographics, interests, intent, persona |
| Audience claim to the buyer | None | Implied from topic | Explicit, field-by-field, reproducible |
| Cookieless inventory | Included but undifferentiated | Included, audience unknown | Same audience profile as cookie-bearing inventory |
| Long-tail coverage | Indiscriminate | Limited by category data | 102M domains, filterable head to tail |
| Maintenance | Blocklist churn | Manual list reviews | Re-run the archived query at each quarterly refresh |
Both, because curation sits between them. SSP and curation-platform teams build sellable packages; agency and in-house programmatic teams specify what they want assembled or build their own allow-lists. The package definition language — coded attributes from fixed vocabularies — is the same on either side of the deal.
Quality control is a selection-time property, not a bid-time decision. Combine confidence floors (medium or high), audience-type restrictions, corpus-tier membership and IAB categorization for suitability. Low-evidence or unsuitable domains never enter the inclusion list — which is where curation quality is actually decided.
The package is defined as a query over versioned vocabularies, and the database refreshes quarterly ($190/quarter for the top-100k tier, $590/quarter for the top-1M tier). Re-running the archived query produces the updated inclusion list; fixed v1.0 codes keep the query’s meaning stable between refreshes.
Yes — that is the point. Every claim in the deal one-sheet corresponds to a field and value a buyer can inspect, either in a shared extract or interactively in the demo dashboard. Claims like “family households with baby-product intent” become checkable statements, not marketing copy.
Build the target domain list a package starts from.
Declare the same audience meaning in the bid request as SDA segments.
Score packaged domains for brand fit, beyond the blocklist.
Curate offsite inventory that matches retail audiences.
Filter the data live in the demo dashboard, or buy the top-1M domain file with instant download and have an inclusion list in front of a buyer this week.
Open the audience demo See database pricing