Vol. III · §3.2 · audited apr '26
Methodology.
§1 · Why this exists
Most "best affiliate program" lists rank by the headline commission rate. That number lies. A 70% commission on a $20 SaaS with a 7-day cookie and 90-day clawback pays a creator less than a 15% commission on a $400 product with a 6-month cookie and net-30 terms. The headline is the marketing brochure, not the math.
FintechPays grades by 12-month true-EPC — the realistic dollars-per-click we expect a competent creator to earn, after every leak. Six factors. Every value editor-set per program. Every program inspectable on its own review page.
§2 · The formula
true_epc_12mo = base_payout
× cookie_decay
× attribution_factor
× reliability_factor
× conversion_rate_estimate
× (1 / payment_threshold_friction)
§3 · Rubric weights
Above the per-factor scoring, the composite weighting reflects how much each domain shapes the realistic EPC vs. the marketing claim:
| FACTOR | WEIGHT |
|---|---|
| commission economics | 30% |
| payout reliability | 25% |
| cookie + attribution | 15% |
| e-e-a-t / network access | 15% |
| audience fit | 10% |
| regulator coverage | 5% |
§4 · Factor definitions
base_payout (USD)
Projected USD revenue per signed-up referral over 12 months. For recurring programs we project 12 months of typical subscription value × commission %. For one-time CPAs it is the CPA itself. Programs that hide the math behind tier ladders ("up to 70%") get the realistic tier, not the ceiling.
cookie_decay (0.10–0.95)
Programs with a 365-day cookie capture nearly every referral; 30-day cookies miss high-consideration buyers; 24-hour cookies miss most of them.
- 365-day cookie → 0.95
- 180-day → 0.85
- 90-day → 0.75
- 60-day → 0.65
- 30-day → 0.55
- 14-day → 0.40
- 24–48h → 0.20
- last-click only → 0.10
attribution_factor (0.25–1.00)
Does the program credit the affiliate honestly, or does its own retargeting overwrite the cookie? Full last-click + open attribution scores 1.00; clawback risk drops to 0.85; shared with own-funnel marketing drops to 0.50; known cookie-overwrite on retargeting hits 0.25.
reliability_factor (0.20–1.00)
Does the program actually pay? Defaults to 1.00, degraded by Trustpilot non-payment threads (−0.20 for 10+ verified complaints), Reddit first-hand reports, defunct-ancestor brand history (−0.40), Net-60+ terms, $500+ minimum thresholds. Floors at 0.20 — programs below 0.40 ship with caution flags; below 0.30 with do-not-promote editor flags; below 0.20 move to the defunct archive.
conversion_rate_estimate (0.01–0.20)
Niche-anchored. Prop trading and crypto-exchange land 0.05–0.15 (high-intent, money-on-the-line). BNPL hits 0.10–0.20 (impulse, low-friction). Robo-advisor sits 0.02–0.05 (low-intent, KYC friction). Editor adjusts per program based on landing-page quality and country-of-traffic match.
payment_threshold_friction (1.0–2.0)
High minimum payouts effectively delay cash and lower the true-EPC. $50–100 minimum → 1.0. $200–500 → 1.3. $500–1000 → 1.5. $1000+ → 1.7. Annual-cycle payouts → 2.0. Applied as × (1 / payment_threshold_friction).
§5 · Display grades are relative, not absolute
Within a (niche × market) cell, the program with the highest true_epc_12mo always scores 100. The rest scale linearly against that ceiling. A B+ in prop-trading × US means "good, but not the best in this market" — it does NOT mean "objectively medium." Cross-cell comparisons aren't apples-to-apples; every program's review page surfaces the absolute EPC alongside the relative grade.
§6 · Per-program audit trail
Every program's review page renders its full factor configuration as a JSON-style breakdown — the six values, the computed EPC, the editor notes documenting where each non-default factor came from. No factor adjustment ships without a documented editor_note. Audit trail is non-negotiable.
§7 · Re-verification cadence
Every program is re-verified on a 90-day cycle. If a payout reliability complaint surfaces, the reliability_factor drops in the next sweep. The audit-trail log on each review page records every change with date stamps. The next sweep is scheduled 2026-08-12.
§8 · v1 → v2 calibration
The v1 formula ships with editor-set factor ranges; the first 2 weeks of live affiliate-network payout data will calibrate v2. Expected v2 changes: (a) refine the conversion_rate_estimate per niche based on actual outbound-CTR vs program-side-conversion data; (b) re-baseline some reliability_factor defaults; (c) possibly simplify payment_threshold_friction to a flat −10% adjustment for thresholds above $500 if the noise drowns the effect.
† Authored by Maren Holst · fact-checked by Asha Devi · audited Apr '26 · methodology version v3.2