Every affiliate network runs on the same two judgments: is this publisher site worth approving, and which programs does it fit? Today both mostly rest on self-description and manual review. Joining each affiliate domain against a 102M-domain database of pre-computed audience segmentation replaces guesswork with a structured profile — demographics, interests, purchase-intent segments, personas and banded confidence, every value an enumerated code from fixed vocabularies aligned with the IAB Audience Taxonomy 1.1. The result is faster vetting, sharper program matching, and an advertiser-facing directory that describes publishers by audience rather than by category checkbox — with no cookies and no PII anywhere in the pipeline.
The same domain-keyed dataset serves publisher development, advertiser success and network quality — because all three argue about the same thing: whose audience is this?
An applicant is a URL and a pitch. The database turns the URL into evidence: measured audience character, b2c/b2b orientation and confidence band, ready before a reviewer opens the site. Profiles that match the claimed niche move fast; profiles that don't get human attention first.
An advertiser's program has a target audience; every publisher domain has a measured one. Matching becomes a code-overlap query — interests, PI.* intent segments, income bands, personas — instead of a category dropdown that lumps every "lifestyle" site together. The advertiser-side version of this workflow is affiliate site selection.
Audience-fit is a leading indicator of program health: partners whose audience matches the offer convert on merit. Recurring divergence between a site's measured audience and the programs it drives volume into is a consistency flag worth investigating — a structured input to quality reviews, not a verdict.
One join key — the normalized registrable domain (eTLD+1) — connects your publisher records to the dataset. Everything else is ordinary batch and API work.
Batch-join every approved and pending affiliate domain against the database. New applicants outside the file are profiled live via the real-time API — same codes, same confidence bands.
Compare each applicant's measured profile — interests, intent, audience_type — with its claimed niche. Consistent profiles auto-advance; mismatches route to manual review.
Express each program's target as code conditions and rank publishers by overlap. Surface the ranked matches to advertisers as recommendations, gated by confidence.
Quarterly refresh files update every profile in place. Diffing releases shows which publisher sites changed audience character — an early signal for both new opportunities and quality drift.
The attribute set is the same on every row: 8 age brackets, 5-point gender skew, 6 income bands, 7 education levels, 14 life stages, household and employment enums, 29 interest groups with 285 sub-interests (INT.*), 34 purchase-intent groups with 283 segments (PI.*), B2B firmographics and 1,667 personas — all enumerated in the taxonomy reference, all versioned so your matching rules keep meaning exactly the same thing across refreshes.
An outdoor-gear review site applies to the network, requesting three programs: camping equipment, hiking apparel and a premium credit card. Its domain profile:
trailgear-reviews.example
affiliate application #48112 · vocab v1.0
25_3435_44male_leanupper_middlesuburbanINT.travel.campingINT.travel.adventure_travelPI.sporting_goods.outdoor_recreation_equipmentPI.clothing_accessories.clothingb2cThe decision writes itself: approve for the camping-equipment and hiking-apparel programs — measured intent segments overlap the offers directly, at high confidence. The premium-card program shows no audience support in the profile; rather than a blanket approval, the network holds that request for the advertiser's own criteria. Vetting time drops, and the two approvals start with genuine audience fit.
Most network directories still describe publishers with a one-of-N category. Here is what changes when the description is a coded audience profile.
| Network task | With self-declared categories | With domain-level audience data |
|---|---|---|
| Application review | Reviewer reads the site and takes the niche on trust | Independent profile ready at triage; claim-vs-measurement mismatches routed to humans |
| Publisher directory | "Sports & Outdoors" contains thousands of unlike sites | Advertisers filter by INT.*/PI.* codes, income band, persona, b2c/b2b |
| Program recommendations | Category adjacency | Ranked code-overlap between program target and measured audience |
| Quality review | Conversion anomalies investigated cold | Audience-consistency signal adds context before escalation |
| Coverage | Whatever applicants typed in | 102M domains; long-tail applicants profiled via API at apply time |
The advertiser-side workflow: choosing affiliate partners by measured audience fit — useful for your advertiser-success playbooks.
Scoring domains against a brand's audience definition — the same overlap logic your program matching runs at network scale.
The sibling problem: describing and packaging publisher inventory by audience for a network's demand side.
Agencies evaluating partners on behalf of clients run a related process — see the agencies page.
At application time, an affiliate is a URL and a self-description. Joining the domain against the audience database attaches a structured, independently derived profile: demographics, interests, purchase-intent segments, personas and a banded confidence score. Reviewers see what the site's content actually indicates about its audience — and can compare it to the niche the applicant claims. A consistent profile speeds approval; a mismatch is a flag for closer manual review before the affiliate ever touches a program.
No, and we don't claim it can. The attributes describe the audience a domain's content is built for — they say nothing about traffic sourcing, cookie stuffing or conversion manipulation. What the data does provide is a consistency signal: when a site's measured audience character diverges sharply from its claimed niche or its referred-conversion pattern, that inconsistency is worth a human look. Treat it as one structured input to your existing quality process, not a fraud verdict.
Under a data licensing agreement, the network hosts the dataset and surfaces it in product: an audience panel on each affiliate listing, audience filters in the advertiser-facing publisher directory, and match conditions in program recommendations. Licenses are quoted individually by corpus size (5M up to the full 102M domains), refresh cadence and product surface; custom licensing starts from $15,000/year — contact us for a quote. Self-serve tiers (Top 100k at $490, Top 1M at $1,990, instant download) cover internal evaluation and ops use; see the pricing page.
That is precisely why the corpus is 102 million domains: affiliate publishers live far beyond any top-1M list. Networks typically license a slice sized to their publisher base and route the remainder through the real-time API, which returns the same coded attributes for any domain or URL on demand — API plans start with the same tiers as our categorization API, e.g. Pro at $99/month for 10,000 credits. New applicants can therefore be profiled at the moment they apply, whether or not they appear in the batch file.
Look up any affiliate domain in the demo dashboard, then run a batch join against your directory with a self-serve tier — the Top 1M file is an instant card-checkout download.