An affiliate site converts when its audience matches the offer — not when its traffic number is big. Profile any of 102 million domains on coded demographics, interests and purchase intent, then compare each candidate to your offer's target before the first commission is paid.
Recruitment still runs on proxies that never describe the thing that determines conversion: who the audience is.
Two personal-finance blogs with identical traffic perform completely differently on a credit-card offer. One reaches affluent investors; the other reaches readers in debt. The category says “finance” for both.
Wrong-audience partners depress aggregate conversion rates, distort the numbers used to set commissions, and consume recruiting time better spent on the right sites.
In Safari, Firefox and iOS (~40%+ of the web), third-party cookies are already blocked. Noisier post-click attribution makes pre-selection on audience fit more valuable, not less.
Content-derived profiles put the match on paper before the relationship starts. “Does this site fit this offer?” becomes a comparison of two coded records with the same attributes.
Four steps, whether you are vetting one inbound applicant or scoring a network's entire publisher list. The full attribute vocabulary is on the audience segmentation taxonomy page.
Translate the offer into the same vocabulary the domains are described in: target age brackets, income band, life stage, and above all the purchase-intent segments that predict conversion — e.g. PI.finance_insurance.credit_cards for a card offer.
Look up applicants and prospects in the pre-computed database, or profile specific sections and landing pages with the real-time API when a site's verticals differ from its homepage. Every candidate returns the same field set.
Compare offer codes to candidate codes: intent-segment overlap first, demographics and life stage second, confidence bands as the tiebreaker. A simple rule — require the primary intent segment at medium-or-high confidence — filters most mismatches immediately.
Rank candidates into tiers: priority recruits, acceptable with adjusted terms, decline. The coded evidence doubles as the internal justification — and as the outreach angle when courting the priority sites.
Illustrative scenario: a cashback credit-card program is vetting an applicant blog. The offer's coded target sits on the left; the candidate's content-derived profile on the right.
PI.finance_insurance.credit_cards age_bracket: 25_34, 35_44 income_level: middle life_stage: family_young_children
Recruit. Primary intent at high confidence (PI.finance_insurance.credit_cards), demographics align. The Debt Repair signal favours the no-annual-fee angle. A rewards-travel card would be a decline — same blog, same traffic, wrong offer.
Fit is a property of the pair (site, offer), not of the site alone. Coded profiles let you evaluate every pair explicitly — and re-evaluate cheaply when the offer lineup changes.
The usual signals aren't wrong — they just answer a different question. Traffic tells you how many; the audience profile tells you who.
| Vetting signal | What it tells you | What it misses | Role in an audience-first workflow |
|---|---|---|---|
| Traffic estimates | Order-of-magnitude reach | Who the visitors are; whether they convert on your offer | Volume sanity check after fit is established |
| Domain authority / SEO metrics | Search competitiveness | Audience composition entirely | Durability signal for long-term partners |
| Network category labels | Broad topic ("Finance", "Lifestyle") | The difference between investors and debt-repairers | Coarse pre-filter only |
| Manual site review | Editorial quality, brand safety feel | Doesn't scale; inconsistent between reviewers | Final check on priority recruits |
| Coded audience profile | Demographics, interests, intent segments, personas with confidence bands | Not a visitor count — pair it with reach data | The core fit decision, uniform across all candidates |
For network-scale scoring — thousands of publisher applications or auditing a partner roster — use the pre-computed database: top 100k at $490, top 1M at $1,990 with instant checkout, or a vertical slice from $190.
Code your offer's target in the same vocabulary — primarily the purchase-intent segment predicting conversion, plus demographics and life stage. Pull the candidate's profile and compare. Require the primary PI.* segment at medium-or-high confidence. Try a single lookup in the live demo in under a minute.
Yes. The pre-computed database covers 102M domains with identical coded fields. Scoring a publisher list is a join on domain plus a rules pass — no per-site research. License the top-1M file ($1,990) or a vertical slice ($190–$490), join to your roster, and re-score when offers change. Quarterly refreshes keep profiles current.
Use the real-time API for page-level profiles. Many affiliate sites are multi-vertical, and the domain-level average can understate the fit of a specific section. Profiling placement URLs gives the precise view. API plans start at $99/month for 10,000 credits. See the API documentation.
No — it front-loads it. Audience-fit vetting predicts which partners to recruit before performance data exists, especially for new partners and cookieless environments where attribution is noisier. Once a partner is live, conversion data is ground truth; profiles then explain why performance differs and which lookalikes to recruit next.
The same profile-and-match pattern applies wherever you choose media by audience.
Paste a candidate site into the live demo and read its coded audience profile — or license the database and score your whole recruitment pipeline in one pass.