Data Brain Performance Decomposition Framework

Document Type: Framework
Status: Active
Version: v1.0
Authority: Data Brain
Applies To: Data Brain, Affiliate Brain, Experimentation Brain, Research Brain, Finance Brain, HeadOffice
Parent: Data Brain Canon
Last Reviewed: 2026-04-25


Purpose

The Data Brain Performance Decomposition Framework defines how MWMS breaks performance changes into their underlying causes.

Its purpose is to prevent MWMS from misreading surface-level performance movement.

Revenue, conversions, or traffic may rise or fall for many reasons.

This framework ensures MWMS identifies what is actually driving performance change before decisions are made.

This framework is informed by the CXL transactional analysis concept of breaking business performance into meaningful components such as customer groups, product classes, channel behaviour, discount behaviour, and time-based comparisons.


Core Principle

Performance must be decomposed before it is interpreted.

A result is not enough.

MWMS must understand:

• what changed
• where it changed
• when it changed
• which segment caused it
• whether the change is healthy or unstable


Definition

Performance decomposition is the process of breaking a top-level result into its contributing parts.

Instead of asking:

“Did performance improve?”

MWMS asks:

“Which part of the system caused the change?”


Core Question

This framework answers:

👉 What is really driving the performance change?


Decomposition Layers


1. Customer Layer

Breaks performance into customer type.

Segments may include:

• new customers
• returning customers
• reactivated customers
• loyal customers
• low-quality customers
• high-value customers

Purpose:

• identify whether growth is coming from strong or weak customer groups
• prevent new customer volume from masking weak retention
• detect customer quality problems


2. Offer Layer

Breaks performance into offer-level contribution.

Segments may include:

• offer
• product
• vendor
• payout type
• offer category
• offer age
• offer quality tier

Purpose:

• identify which offers are driving or weakening results
• separate traffic problems from offer problems
• detect offer decay


3. Channel Layer

Breaks performance into traffic and marketing source.

Segments may include:

• Google Ads
• YouTube
• organic search
• email
• social
• affiliate/referral
• direct traffic

Purpose:

• identify whether performance change is caused by channel quality
• avoid blaming offer quality when traffic source is the issue
• avoid blaming traffic when the offer is weak


4. Creative / Angle Layer

Breaks performance into message-level contribution.

Segments may include:

• hook
• angle
• claim structure
• trust signal
• emotional trigger
• creative format

Purpose:

• identify which messages create durable behaviour
• detect weak or shallow engagement
• connect creative output to customer quality


5. Funnel Layer

Breaks performance into journey stage.

Segments may include:

• ad click
• landing page visit
• VSL click
• form start
• purchase
• lead submission
• post-conversion behaviour

Purpose:

• identify where friction appears
• separate attention problems from conversion problems
• detect funnel-stage weakness


6. Time Layer

Breaks performance into time-based changes.

Segments may include:

• day
• week
• month
• launch period
• pre-change period
• post-change period

Purpose:

• detect when performance changed
• connect changes to system events
• identify trend shifts


7. Price / Promotion Layer

Breaks performance into full-value versus incentive-driven behaviour.

Segments may include:

• full-price behaviour
• discount-driven behaviour
• promotion-driven behaviour
• low-margin behaviour

Purpose:

• detect fake performance gains
• identify customer behaviour degradation
• prevent revenue growth from hiding profit weakness


Diagnostic Use

This framework should be used when:

• performance changes unexpectedly
• campaigns appear to improve but profit does not
• revenue grows while quality declines
• offer performance weakens
• customer behaviour shifts
• scaling decisions are being considered
• HeadOffice needs a clear explanation


Performance Decomposition Workflow


Step 1 — Identify Top-Level Change

Define the observed movement.

Examples:

• revenue increased
• conversion rate dropped
• customer quality declined
• offer stopped working
• campaign performance weakened


Step 2 — Break Into Layers

Decompose the result across:

• customer
• offer
• channel
• creative
• funnel
• time
• price / promotion


Step 3 — Locate the Driver

Identify which layer shows the strongest explanation.

Example:

Revenue increased because new customer volume rose, but repeat behaviour declined.


Step 4 — Validate Against Data Trust

Data Brain must confirm:

• data is clean
• segmentation is valid
• attribution is reliable enough
• signal integrity is acceptable


Step 5 — Classify Cause

Classify the likely cause as:

• customer quality issue
• offer issue
• traffic issue
• creative issue
• funnel issue
• pricing / promotion issue
• measurement issue
• external factor


Step 6 — Route To Correct Brain

Send the finding to the correct owner.


Cause Ownership Map


Customer Quality Issue

Owner:

• Customer Brain
• Data Brain
• HeadOffice


Offer Issue

Owner:

• Affiliate Brain
• Offer Brain
• Research Brain


Traffic Issue

Owner:

• Ads Brain


Creative Issue

Owner:

• Creative Brain
• Ads Brain


Funnel Issue

Owner:

• Conversion Brain


Pricing / Promotion Issue

Owner:

• Finance Brain
• Offer Brain
• HeadOffice


Measurement Issue

Owner:

• Data Brain


External Factor

Owner:

• Research Brain
• HeadOffice


Decision Rules

MWMS must not act on top-level performance alone.

Before making a decision, MWMS must identify:

• primary driver
• supporting driver
• confidence level
• affected Brain
• recommended next action

If no driver can be identified:

→ decision must be delayed or classified as uncertain.


Scaling Rule

Scaling must not occur if performance improvement is caused by:

• weak customer quality
• discount dependency
• temporary promotion lift
• unstable traffic mix
• measurement error
• unvalidated attribution

Scaling may proceed only if:

• decomposition shows healthy drivers
• data trust is acceptable
• customer quality is stable or improving
• profit logic is valid


Cross Brain Use


Data Brain

Owns decomposition structure, validation, and signal trust.


Affiliate Brain

Uses decomposition to identify whether offer performance is genuinely strong.


Experimentation Brain

Uses decomposition to determine what should be tested next.


Research Brain

Uses decomposition outputs to identify recurring patterns.


Finance Brain

Uses decomposition to protect capital from false performance signals.


HeadOffice

Uses decomposition to understand business truth and assign action.


Relationship To Other Frameworks

This framework connects to:

• Data Brain Customer Quality Tracking Framework
• Data Brain Segmentation Framework
• Data Brain Data Trust Framework
• Data Brain Attribution Reliability Framework
• Experimentation Brain Diagnostic Trigger Framework
• Affiliate Brain Offer Health Monitoring Framework
• MWMS Promotion Impact Framework
• HeadOffice Business Diagnostic Narrative Framework


Failure Modes Prevented

This framework prevents:

• blaming the wrong Brain
• scaling false positives
• missing customer quality decline
• misreading revenue growth as health
• confusing traffic problems with offer problems
• confusing offer problems with funnel problems
• ignoring price or promotion distortion
• acting before understanding cause


Drift Protection

The system must prevent:

• aggregate performance replacing segmented diagnosis
• top-line metrics being treated as complete truth
• hidden segment decay going unnoticed
• recurring causes being missed
• weak explanations being accepted as final


Architectural Intent

Performance Decomposition gives MWMS a diagnostic layer beneath results.

It transforms MWMS from:

“Performance changed”

to:

“Performance changed because this part of the system changed.”

This strengthens MWMS diagnostic intelligence and decision confidence.


Final Rule

No major MWMS decision should be made from top-level performance alone.

Performance must be decomposed before action.


Change Log

Version: v1.0
Date: 2026-04-25
Author: Data Brain / HeadOffice

Change

Initial creation of Data Brain Performance Decomposition Framework based on transactional analysis intelligence extracted from CXL course material.

Change Impact Declaration

Pages Created:
Data Brain Performance Decomposition Framework

Pages Updated:
None

Pages Deprecated:
None

Registries Requiring Update:
Data Brain Architecture
MWMS Architecture Registry

Canon Version Update Required:
No

Change Log Entry Required:
Yes


End of Framework