Document Type: Specification
Status: Active
Version: v1.1
Authority: Affiliate Brain
Applies To: Affiliate Brain performance interpretation, downstream outcome-quality review, and closed-loop profitability discipline
Parent: Affiliate Brain Architecture
Last Reviewed: 2026-03-14
Purpose
This document defines the closed-loop feedback layer used to prevent false optimisation signals by linking campaign performance to downstream outcome quality.
Affiliate platforms often report only front-end activity such as:
• clicks
• leads
• initial conversions
But real business truth depends on downstream quality indicators such as:
• refunds
• chargebacks
• upsell acceptance
• retention, where recurring
• net revenue after reversals
This layer exists to ensure that a reported conversion is treated as provisional until outcome quality is validated.
Scope
This specification applies to:
• Affiliate Brain performance interpretation
• downstream validation of campaign quality
• net profitability judgement after reversals
• manual-first logging of buyer-quality indicators
• future closed-loop integration into governance systems
This document defines how Affiliate Brain should interpret performance beyond front-end conversion data.
It does not govern:
• live campaign execution
• bid changes
• creative production
• direct finance approval
• automated network integrations at the current stage
Those remain governed by Affiliate Brain Canon, Finance Brain, Velocity Decision Engine, and future automation layers when activated.
Definition / Rules
Closed-Loop Feedback Layer (Affiliate)
This page defines the closed-loop feedback discipline for affiliate performance review.
Its role is to connect top-level campaign metrics with downstream business outcomes.
Core Principle
Front-end conversion metrics are not final truth.
A campaign may appear profitable while silently failing due to:
• high refund rate
• high chargeback rate
• poor buyer quality
• low upsell capture
• weak retention
Therefore, performance decisions must eventually incorporate outcome-quality signals.
Affiliate Closed-Loop Signals (Primary)
Required outcome-quality signals include:
- Refund Rate
- Chargeback Rate
- Net Revenue, after reversals
- Upsell Acceptance Rate, where available
- Rebill or Retention Signal, where recurring
These signals help distinguish apparent performance from real performance.
Decision Impact Rules
Closed-loop signals may override top-level ad metrics.
Examples:
• CTR high + CVR high + Refund high → FAIL (buyer quality failure)
• CTR low + CVR high + Refund low → POSSIBLE WIN (creative issue only)
• CPA acceptable + Net Revenue low → HOLD (true profitability failure)
This prevents the system from rewarding surface-level success that does not hold up financially.
Integration with Existing Systems
This layer is intended to feed into:
• Velocity Decision Engine, for scale gating
• Finance Brain, for capital allocation safety
• Offer Intelligence, for offer risk scoring
• Ad Metadata Registry, for ad-level quality tagging
• Two-Hurdle Framework, by upgrading the conversion hurdle to net conversion
These integrations improve portfolio-level discipline.
Current Status
Affiliate closed-loop data access varies by network.
Therefore, this layer is currently defined as:
Manual-first discipline
Automation later
Minimum Implementation (Manual)
For each tested offer, log monthly:
• refund estimate or refund count, if visible
• reversal notes, if visible
• vendor quality warning signals
• recurring retention signals, if visible
• manual net profitability judgement
This is the minimum discipline required until consistent tracking availability is confirmed.
Future Automation (Deferred)
Potential future integrations include:
• network reporting ingestion
• refund event logging into Supabase
• offer-level outcome quality dashboard
• portfolio governance buyer-quality scoring
Execution remains deferred until tracking availability is confirmed.
Activation Status
Current activation state:
Discipline Defined
Automation Deferred
Drift Protection
The system must prevent:
• treating front-end conversions as final truth
• scaling offers without reviewing downstream quality
• ignoring refunds, reversals, or retention weakness
• assuming buyer quality from CTR or CPA alone
Performance interpretation must remain tied to business reality, not surface metrics.
Architectural Intent
Affiliate Performance Systems exists to protect MWMS from false optimisation confidence.
It ensures that campaign success is evaluated not only by early conversion signals but by downstream commercial truth.
Over time, this improves capital protection, offer selection quality, and scaling discipline.
Change Log
Version: v1.1
Date: 2026-03-14
Author: MWMS HeadOffice / Affiliate Brain
Change: Rebuilt page to align with MWMS document standards. Added standardised document header, corrected parent structure, replaced Linked Canon and Last Updated with compliant metadata, introduced Purpose / Scope / Definition / Rules structure, and preserved the original closed-loop feedback logic.
Version: v1.0
Date: 2026-02-26
Author: Affiliate Brain
Change: Initial creation of Affiliate Performance Systems defining the affiliate closed-loop feedback layer, outcome-quality signals, decision impact rules, manual implementation discipline, and future automation direction.
END – AFFILIATE PERFORMANCE SYSTEMS v1.1