How do you score adult affiliate networks?

Six weighted dimensions decide the composite: Offer Breadth 18 percent, EPC Reliability 18 percent, Payout Cadence and Reliability 18 percent, Dashboard UX and Reporting Depth 16 percent, Tracking and Postback Architecture 16 percent, AM Responsiveness and Advanced Tools 14 percent. The three dimensions tied at the top weight (18 percent) cover what publishers actually evaluate before committing meaningful traffic to a network: what brands are routable, what they earn per click, and when the money lands.

Why these six dimensions and not others?

These are the six signals publishers verify before committing meaningful traffic. Offer Breadth answers 'can I monetize my niche.' EPC Reliability answers 'how much do I earn per click.' Payout Cadence answers 'when does the money land.' Dashboard UX answers 'can I optimize what I cannot see.' Tracking Architecture answers 'do my SubIDs survive end-to-end.' AM Responsiveness answers 'who unblocks me when an offer breaks.' Other signals (compliance restrictions, geo coverage, application friction) feed into these six rather than standing as separate dimensions.

Do you have operator accounts on the networks you score?

Yes. This is the only scoring system we ship where first-hand operator access materially changes scoring depth. The lead reviewer (Alexandra Joly via the bestgirlfriend.ai operator account) holds active publisher accounts on the networks we cover. Dashboard UX, payout-cadence reliability, tracking-postback architecture, and AM responsiveness are graded from direct operator experience over a six-month observation window. Public data (offer catalogs, EPC tables, blog posts, AffPaying community ratings, industry press) supplements where direct observation is incomplete.

What is the NIV flag?

NIV means 'we haven't verified this directly.' It marks any sub-criterion we cannot observe ourselves, typically when we don't hold a publisher account on a network we're scoring (we cover CCBill as a merchant gateway, not as a publisher), or when a feature lives behind a tier we haven't unlocked yet (Premium tier on CrakRevenue is volume-gated and we annotate that limit explicitly). NIV-flagged sub-scores show in italics with a footnote citing the fallback source: usually network-published policy pages, AffPaying community ratings, or industry press archives.

How do you test EPC reliability?

We track EPC across a six-month operator-account observation window: published EPC versus measured EPC on our own traffic, network EPC fluctuation versus seasonal patterns, and the gap between auto-approved offer EPC and AM-gated Premium tier EPC. The EPC Reliability dimension penalizes networks whose published numbers consistently outperform measured numbers, networks whose Premium tier traps EPC behind opaque thresholds, and networks who don't disclose geo-block EPC fallback handling (the Chaturbate Texas-Utah-Louisiana redirect pattern is a documented example of compliance-driven EPC erosion).

Why is Dashboard UX weighted only 16 percent, lower than Offer Breadth or Payout?

Because operators learn the dashboard. A clunky dashboard is annoying for the first month and invisible by month six. Offer Breadth and Payout reliability and EPC are structural: they decide whether the network is monetizable at all, and they earn the 18 percent weights. Dashboard UX matters and is graded against named comparables (HasOffers, Cake, Voluum benchmarks), but it does not gate revenue the way the top three dimensions do.

What are the tier labels for affiliate networks?

Composite scores map to seven tier labels: Best in class (9.0+), Excellent (8.0-8.9), Strong (7.0-7.9), Good for the niche (6.0-6.9), Average (5.0-5.9), Below average (4.0-4.9), Avoid (≤3.9). Catalog floor 5.0: any network below 5.0 is excluded from recommendations regardless of how favourable the operator-side relationship is.

Does CrakRevenue paying us influence the score?

No. The CrakRevenue review scores the network at 9.2/10 against the same dimension weights regardless of our affiliate relationship with them. The Score-Lock Framework documented at the top of the methodology overview applies identically here. Specifically: (a) the score-floor 5.0 rule cannot be overridden by commission, (b) the score-lock-at-publish rule prevents silent post-publication score changes, (c) the dimension weights cannot be re-tuned to favour a particular network. If CrakRevenue paid us more tomorrow, the 9.2/10 would still be 9.2/10. The Versus pages (CR vs CCBill / CR vs ExoClick / CR vs TrafficJunky / CR vs JuicyAds) display competitor wins honestly when the data supports them.

Are affiliate-network scores comparable to AI companion or cam site scores?

No. The five scoring systems (AI Companion 8-dim, Cam Site 6-dim, Adult Game 7-dim, Real-Model 6-dim, Affiliate Network 6-dim) measure different product categories with different weights. An AI girlfriend app and an affiliate network are different objects of evaluation. Cross-system comparison renders category-by-category narrative rather than a single composite score; the verdict explicitly says 'depends on operator profile' rather than naming a winner.

How do I report a methodology error?

Email editorial@bestgirlfriend.ai with the URL of the affected page and the specific claim you are contesting, with sources if available. Corrections log at the top of the affected page for sixty days and note in our public correction log at /errata/. Network operators contesting their own published score use the same channel; we don't bury complaints from networks we cover.

Editorial standards

How We Score Adult Affiliate Networks (6-Dim Audit)

Q: What are the absolute red lines for affiliate network scoring?

Two non-negotiable disqualifiers. (1) Any network whose terms permit traffic from sources that violate our absolute red lines (CSAM, non-consensual content, deepfakes of real people, bestiality) is excluded from coverage entirely. (2) Any network that runs pay-to-rank arrangements (soliciting publishers to recommend the network in exchange for EPC lifts or volume bonuses) is disqualified from scoring on integrity grounds. We have not encountered the second pattern in the adult-niche space as of 2026; if we did, we would name it publicly and decline the relationship.

Adult affiliate network methodology: 6 dimensions (offer breadth, EPC, payout cadence, dashboards, tracking, AM), first-hand operator audit, no pay-to-rank.

By Alexandra JolyLast revised May 2026Rubric v1.0

By Alexandra Joly, Senior Editor • Last full retest May 14, 2026 • Adult Affiliate Network Methodology • Methodology overview

Key takeaways

Six weighted dimensions: Offer Breadth 18%, EPC Reliability 18%, Payout Cadence and Reliability 18%, Dashboard UX and Reporting Depth 16%, Tracking and Postback Architecture 16%, AM Responsiveness and Advanced Tools 14%.
First-hand operator audit. The lead reviewer holds active publisher accounts on the networks we cover; dashboard UX, payout reliability, tracking architecture, and AM responsiveness are graded from six months of direct operator experience.
Score-Lock Framework applies. The 9.2/10 we publish for CrakRevenue does not move because CrakRevenue pays us more next quarter. The score-floor 5.0 rule, the score-lock-at-publish rule, and the weight-immutability rule all apply identically here.
Two absolute red lines. A network that permits traffic from CSAM, non-consensual, deepfake-of-real-person, or bestiality sources is disqualified. A network that runs pay-to-rank publisher relationships is disqualified on integrity grounds.
NIV (we haven't verified this directly) flag is mandatory anywhere we could not observe the sub-criterion ourselves. The most common NIV gap on adult affiliate network scoring is Premium-tier EPC where the operator account has not yet cleared the volume threshold.

This is the complete public version of the adult affiliate network methodology used in every network review and every cross-network comparison on the /affiliate-marketing/ hub. The detailed sub-criteria checklists, the per-network workflow, and the earnings-reliability test windows behind this page stay internal; what you're reading is the version readers see, plain-spoken but no less honest.

Three benchmarks shaped the design: Wirecutter's published methodology, the AffPaying community-driven network rating archives, and the FTC 16 CFR Part 255 endorsement-disclosure discipline. The combination yields an adult affiliate network methodology that survives outside scrutiny, including scrutiny from the networks being scored.

Why does an adult affiliate network methodology need to be published?

The space is structurally under-reviewed. Most publisher-facing review sites are themselves operators recommending the networks they personally route to, a conflict of interest that explains why the same five networks recur as "top picks" with marketing-page boilerplate. We score networks against a published methodology with first-hand operator data and disclose our own affiliate relationship at the top of every page. The audit protects readers from the recursive-recommendation conflict and protects us from drifting toward the network that pays us best instead of the network that performs best.

Look. I'll be direct here, because the audience for this page is other operators and they can smell a soft pitch from two blocks away. Most "best adult affiliate program" lists you'll find on the open web are pay-to-rank, full stop. The network that bids highest on the page owner's referral commission gets the #1 spot, and the analytical depth is whatever the writer could pad together in an afternoon. I've read enough of them to be tired of it.

The fix isn't to pretend we don't earn commission too (we do, that's the entire business model). The fix is to lock the dimensions and the weights and the sub-criteria publicly, score against them honestly, and document the workflow so a reader can replicate the audit if they want to call us out on something. That's what this page does. The covered set is the five most-routed networks for adult-vertical publishers: CrakRevenue, ExoClick, TrafficJunky, JuicyAds, CCBill (as merchant gateway). AdsTerra sits in the comparison column on the pillar but doesn't get a full audit yet because we don't operate a publisher account on it.

What are the six dimensions and how are they weighted?

Six dimensions sum to 100 percent: Offer Breadth 18% (catalog × vertical coverage × auto-approved share), EPC Reliability 18% (published vs measured, Premium-tier transparency), Payout Cadence and Reliability 18% (NET window, method options, missed-payout history), Dashboard UX and Reporting Depth 16% (SubID granularity, reporting cube flexibility), Tracking and Postback Architecture 16% (S2S, pixel, attribution window), AM Responsiveness and Advanced Tools 14% (ticket median, promo tool depth).

The three top-weight dimensions (Offer Breadth, EPC Reliability, Payout Cadence) carry 54 percent combined because they're the dimensions that decide whether a network is monetizable at all for a publisher in our space. A network with thin offer breadth on the AI vertical is unmonetizable for an AI girlfriend site no matter how clean its dashboard is. A network with 18-month-stale EPC numbers wastes the publisher's traffic for months before reality lands. A network with chronic NET-60 cadence kills the cash flow that lets a small publisher reinvest into more content.

The two middle dimensions (Dashboard UX, Tracking Architecture) carry 16 percent each because they affect daily operations but operators adapt to them. A clunky dashboard is annoying for the first month and invisible by month six. A postback that lacks one SubID token is a routing exercise, not a structural problem.

The bottom dimension (AM Responsiveness and Advanced Tools) carries 14 percent because it matters at the margin: an AM who unblocks a stuck offer in 48 hours versus 14 days changes the publisher's experience meaningfully but doesn't typically gate revenue.

Adult Affiliate Network Methodology, six dimensions, weights, and test methods. Weights sum to 100% across the six.
Dimension	Weight	Test method
Offer Breadth	18%	Catalog audit: total active offers, vertical coverage (AI / cam / fansite / adult gaming), exclusive offers, geo coverage, auto-approved share at signup.
EPC Reliability	18%	Six-month operator-account observation: published EPC vs measured EPC on operator traffic, EPC fluctuation pattern, Premium-tier gating transparency, geo-block EPC fallback handling.
Payout Cadence and Reliability	18%	Operator-account payout history: cadence (NET-15/30/60 or weekly), minimum payout threshold, method options (wire, Paxum, crypto, ePayService), reliability (missed payouts, support friction on payout queries).
Dashboard UX and Reporting Depth	16%	Operator-account UI walk-through: navigation depth, reporting granularity (per-offer / per-SubID / per-geo / per-device / per-source), real-time vs pull-cadence data, exportability, mobile usability. Benchmarked against HasOffers, Cake, Voluum norms.
Tracking and Postback Architecture	16%	S2S postback support, pixel-based tracking option, SubID token count (5+ tokens is the table-stakes minimum for adult-niche granularity), server-side geo-fallback routing, attribution window, postback success rate observed on operator traffic.
AM Responsiveness and Advanced Tools	14%	Affiliate Manager ticket response median (operator-account observation), advanced tool set (Smartlinks, Cam Widgets, Back Offers, Embedded Games, Cam Models API), brand-direct unlock cadence on AM-gated Premium offers.

How does operator-account access change scoring depth?

This is the only scoring system we ship where the lead reviewer's direct operator experience is the dominant source of evidence. On AI companion, cam, adult-game, and real-creator reviews, $0 editorial spend is a deliberate constraint we adopt to remove conflict-of-interest framing. On adult affiliate network scoring, the opposite holds: without a publisher account, dashboard UX, payout reliability, tracking architecture, and AM responsiveness are unscoreable. We disclose the operator relationship explicitly and the Score-Lock Framework protects against commission-driven drift.

The lead reviewer (me, Alexandra Joly, via the bestgirlfriend.ai operator account) holds active publisher accounts on CrakRevenue (primary partner), ExoClick, TrafficJunky, and JuicyAds. CCBill is covered as a merchant-gateway service: we don't operate a CCBill publisher relationship, because CCBill isn't a publisher network. It's a payment processor for adult merchants, and we score it against the dimensions where the comparison makes sense (merchant-side billing reliability, geo coverage, integration friction).

The audit window is six months minimum. Single-incident reports don't move the score; sustained patterns do. EPC volatility tied to seasonal patterns (Q4 holiday surge, January retention drops, June lull) doesn't penalize the network. Volatility tied to silent offer-payout changes does. The Q1 Multi-CPA scrub-rate surprise I documented in the pillar is exactly the kind of pattern the operator window catches, and that a marketing-page reviewer site can never see, because they've never had a payment processed against their account.

Most reviewers in this space won't tell you which networks they actually operate on. The line is usually "we tested extensively" with no specifics. We name the four networks I hold accounts on, and we name the one (CCBill) where the audit is merchant-side only. That asymmetry is the editorial discipline; pretending we hold accounts on all twelve adult networks would be a lie.

What does the NIV flag mean and when do we use it?

NIV means "we haven't verified this directly." It marks any sub-criterion we cannot observe ourselves. The two most common patterns: (1) Premium-tier EPC where the operator account hasn't yet cleared the volume threshold; (2) feature surfaces that vary per geo or per vertical where our operator traffic doesn't exercise all variants. Every NIV-flagged sub-score on a published page carries a footnote naming the gap and citing the fallback source, typically network-published policy pages, AffPaying community ratings (≥5 independent operator reports), or industry-press archives.

NIV doesn't stamp the composite. A network that scores 8.5 on Offer Breadth from first-hand audit and 7.5 NIV on AM-gated Premium tier EPC publishes the composite truthfully with the NIV footnote attached. Hiding a dimension behind NIV is not equivalent to discarding it from scoring, the dimension still counts toward the composite at the fallback-source figure, the footnote just disambiguates how we arrived at the number.

The honesty rule here matters more than it looks. A reviewer who never flags NIV is a reviewer who's either lying about coverage or has a 100-account operator stack nobody else can match. We're in the first category if we don't use NIV; that's why the flag exists.

What do the tier labels mean?

Seven tier labels map composite-score bands. Best in class (9.0+) flagship recommendation; Excellent (8.0-8.9) strong with surfaced weaknesses; Strong (7.0-7.9) secondary recommendation; Good for the niche (6.0-6.9) vertical-specific only; Average (5.0-5.9) niche-only with caveats; Below average (4.0-4.9) not recommended; Avoid (≤3.9) excluded from catalog. Catalog floor 5.0: any network scoring below 5.0 is excluded from recommendations regardless of operator-side incentives. Non-negotiable.

Adult affiliate network tier labels mapped to composite-score bands.
Composite band	Tier label	What it means
9.0 – 10.0	Best in class	Catalog leader on the dominant dimension(s); approved across multiple verticals; flagship recommendation for operators in our space.
8.0 – 8.9	Excellent	Strong across most dimensions with one or two real weaknesses surfaced honestly; recommended for specific operator profiles.
7.0 – 7.9	Strong	Solid on the dimensions that matter most for the specific vertical it serves; secondary recommendation alongside a Best-in-class anchor.
6.0 – 6.9	Good for the niche	Niche-specific recommendation only, typically a network with deep coverage on one vertical but thin breadth elsewhere.
5.0 – 5.9	Average	Niche-only recommendation with caveats disclosed; better fits exist on most operator profiles.
4.0 – 4.9	Below average	Not recommended; structural weakness on at least two of the top-weighted dimensions.
≤ 3.9	Avoid	Does not meet the quality bar; excluded from the catalog regardless of operator-side incentives.

Catalog floor 5.0. Any network scoring below 5.0 is excluded from recommendations regardless of how favourable the operator-side relationship is. This rule supersedes every commercial consideration.

What are the absolute red lines for affiliate network scoring?

Two non-negotiable disqualifiers. (1) Any network whose published terms permit traffic from sources violating our absolute red lines (CSAM, non-consensual content involving real people, deepfakes of real people without consent, bestiality) is excluded from coverage entirely. (2) Any network that solicits publishers to recommend the network in exchange for EPC lifts, volume bonuses, or off-scoring financial considerations is disqualified on integrity grounds.

How does a network actually get scored? The workflow

Ten steps per network before a review publishes. Operator account active → six-month observation window accumulates → dimension scoring against sub-criteria → EPC reliability test → payout reliability test → tracking architecture test → AM responsiveness test → cross-vertical comparison vs peer set → editorial review → lock and publish. Composite locked at publish per the Score-Lock Framework; subsequent changes go into the version history with re-test evidence.

1
Operator account active
Lead reviewer holds a live publisher account on the network being scored (with the CCBill exception noted above as a merchant-gateway audit).
2
Six-month observation window
Direct operator-account data accumulates across at least 26 weeks of routed traffic before the score is locked.
3
Dimension scoring against published sub-criteria
Each of the six dimensions scored 1-10 against the documented sub-criteria, with primary-source citations or NIV flag attached at the sub-criterion level.
4
EPC reliability test
Published EPC versus measured EPC across the observation window, with volatility patterns annotated. AM-gated Premium tiers flagged when applicable.
5
Payout reliability test
Cadence verification, method-option testing, missed-payout incident review across the observation window.
6
Tracking architecture test
S2S postback success rate, SubID token survival, geo-fallback routing on geo-blocked offers, attribution window verification.
7
AM responsiveness test
Sample tickets across the observation window; response-median and resolution-median measured. Brand-direct unlock cadence on AM-gated Premium offers tracked.
8
Cross-vertical comparison
Network scored against its peer set on the same dimensions; versus pages render head-to-head verdicts where the data supports them.
9
Editorial review
The editorial team reviews the page before publish; disagreements greater than 1 point on any dimension are resolved before the page goes live.
10
Lock and publish
Composite score locked at publish per the Score-Lock Framework; subsequent score changes documented in the page's version history with re-test evidence.

How often do scores get re-tested?

EPC Reliability and Payout Cadence re-test every 3 months (network policy changes happen quarterly). Offer Breadth re-tests every 6 months unless a major vertical addition or removal triggers an early re-test within 30 days. Dashboard UX, Tracking Architecture, and AM Responsiveness re-test every 6 months. Major events (a network adds a vertical, a tracker swap, a payment-rail change, a major compliance shift, a regulatory action) force an early re-test on the affected dimension within 14 days.

The asymmetric cadence is deliberate. EPC and Payout move on quarterly network policy cycles, so a quarterly cadence catches the drift early. Offer Breadth moves on annual or semi-annual planning cycles, so semi-annual catches the structural shifts without churning the page. Dashboard / Tracking / AM rarely change without an announcement, so the same six-month cycle works, with the 14-day exception covering any sudden change. The publisher reading this can trust the dates on each dimension's freshness footnote; they're not decorative.

Version history

Version	Date	Change
v1.0	2026-05-14	Initial public lock. 6 dimensions documented as applied across the 8 affiliate-marketing hub pages shipped 2026-05-14. Operator-account audit protocol disclosed publicly.

Sources

[Source: FTC 16 CFR Part 255, Guides Concerning the Use of Endorsements and Testimonials · verified 2026-05-26]
[Source: AffPaying, community-driven affiliate-network rating archive · verified 2026-05-26]
[Source: HasOffers / TUNE, dashboard benchmark for B2B tracker UX (CrakRevenue underlying platform) · verified 2026-05-26]

Methodology overview, the five scoring systems applied across the catalog
Affiliate marketing hub, the pillar with the network recommendations this scoring produces
CrakRevenue review, full 9.2/10 audit against the six dimensions
About bestgirlfriend.ai and Alexandra Joly, masthead, editorial team, persona
Editorial process, how briefs are commissioned, reviewed, and published
Affiliate disclosure, Score-Lock Framework + affiliate-payout independence rules
Errata, public correction log

Last verified May 26, 2026 · See errata log for any post-publish corrections · Editor: Alexandra Joly · Methodology overview · Editorial process · Affiliate disclosure

Why does an adult affiliate network methodology need to be published?

What are the six dimensions and how are they weighted?

How does operator-account access change scoring depth?

What does the NIV flag mean and when do we use it?

What do the tier labels mean?

What are the absolute red lines for affiliate network scoring?

How does a network actually get scored? The workflow

Operator account active

Six-month observation window

Dimension scoring against published sub-criteria

EPC reliability test

Payout reliability test

Tracking architecture test

AM responsiveness test

Cross-vertical comparison

Editorial review

Lock and publish

How often do scores get re-tested?

Version history

Sources

Related reading