MWMS MCR

Experimentation Brain Personalization Testing Framework

Document Type: Framework
Status: Canon
Authority: HeadOffice
Applies To: Experimentation Brain, Conversion Brain, Ecommerce Brain, Ads Brain, Data Brain, Customer Brain, Affiliate Brain, AIBS Brain
Parent: Experimentation Brain Canon
Version: v1.0
Last Reviewed: 2026-05-03

Purpose

The Personalization Testing Framework defines how MWMS designs, executes, and evaluates:

  • A/B testing
  • rules-based personalization
  • predictive personalization

This framework ensures that personalization is not random, unstructured, or based on assumption.

It creates a unified system for:

  • deciding what type of testing to use
  • managing multiple experiences
  • learning from results
  • scaling winning strategies

Scope

This framework applies to:

  • website personalization
  • funnel optimization
  • landing page testing
  • ecommerce personalization
  • ad-to-page experience alignment
  • AIBS user experience systems

It governs:

  • testing method selection
  • experience creation
  • variation management
  • traffic allocation
  • learning cycles

Definition

Personalization Testing

The structured process of testing different user experiences to determine which variation performs best for a given audience or context.

Core Principle

Not all users should see the same experience.

But:

Not all personalization should be permanent or assumed.

Every experience must be:

→ tested
→ validated
→ measured

Personalization Methods

1. A/B Testing (Single Winner Model)

Definition

Test two or more variations to determine one best-performing experience.

Characteristics

  • one winner selected
  • shown to all users after test
  • fixed experience

Best Used When

  • testing major changes
  • testing messaging direction
  • testing layout or structure
  • testing offer differences

Strengths

  • clear results
  • simple decision-making
  • strong statistical clarity

Limitations

  • slow learning
  • requires traffic
  • assumes one experience fits all users

2. Rules Based Personalization (Segment Model)

Definition

Different experiences are shown based on predefined user segments.

Structure

Each rule defines:

  • who sees it (segment)
  • what they see (experience)
  • where it applies (context/page)

Best Used When

  • segments behave differently
  • clear audience differences exist
  • known intent groups exist

Examples

  • mobile vs desktop
  • new vs returning users
  • location-based messaging
  • traffic source variations

Strengths

  • targeted experiences
  • higher relevance
  • controllable logic

Limitations

  • requires segmentation accuracy
  • rules can become complex
  • difficult to scale manually

3. Predictive Personalization (Dynamic Model)

Definition

System dynamically allocates traffic to multiple experiences based on performance.

Characteristics

  • multiple experiences active
  • no single permanent winner
  • traffic shifts over time

Best Used When

  • high traffic environments
  • multiple viable experiences
  • user behaviour varies significantly
  • continuous optimization required

Strengths

  • adapts over time
  • captures more total performance
  • optimizes per user type

Limitations

  • requires data volume
  • more complex implementation
  • harder to interpret

Method Selection Framework

Decision Logic

Use A/B Testing When:

  • one experience should win
  • major directional decision is needed
  • traffic is limited
  • clarity is required

Use Rules Based Personalization When:

  • clear segments exist
  • behavior differs by group
  • messaging needs to adapt

Use Predictive Personalization When:

  • multiple strong variations exist
  • behavior is unpredictable
  • traffic is high
  • continuous optimization is desired

MWMS Default Approach

A/B Test → Identify winners
→ Apply Rules → Segment users
→ Scale with Predictive Personalization

Experience Design Structure

Each test must define:

Campaign

The high-level test objective.

Experience

The version of the user journey.

Variation

Specific differences inside the experience.

Hypothesis

What is being tested and why.

Example

Campaign: Improve conversion rate
Experience A: Current page
Experience B: New messaging
Variation B1: Headline change
Variation B2: CTA change

Testing Workflow

Step 1 — Ideation

  • identify opportunity
  • define hypothesis
  • select optimization target

Step 2 — Creation

  • build variations
  • define segments (if needed)
  • prepare experiences

Step 3 — Implementation

  • launch test
  • assign traffic
  • validate tracking

Step 4 — Measurement

  • collect data
  • monitor performance
  • ensure validity

Step 5 — Learning

  • analyse results
  • identify insights
  • validate conclusions

Step 6 — Iteration

  • refine variations
  • expand winning ideas
  • launch next test

Traffic Allocation Rules

A/B Testing

  • evenly split traffic
  • ensure statistical validity

Rules Based

  • assign traffic based on segment logic

Predictive

  • dynamically allocate based on performance

Measurement Requirements

Each test must track:

  • primary optimization target
  • secondary metrics
  • segment performance
  • device performance
  • traffic source performance

Results Analysis

Results must be analysed by:

  • audience segment
  • variation
  • device
  • channel
  • time

Insight Rule

Every test must produce:

  • what worked
  • what failed
  • why it happened
  • what to test next

Common Mistakes

1. Using A/B Testing For Everything

  • ignores segment differences

2. Overcomplicating Rules

  • creates management complexity

3. Misusing Predictive Systems

  • without enough data

4. Not Validating Results

  • false conclusions

5. No Learning Loop

  • repeating mistakes

Governance Role

This framework ensures:

  • consistent testing methodology
  • controlled personalization
  • unified experimentation logic
  • scalable learning system

Relationship To Other MWMS Standards

  • Experimentation Brain Optimization Target Selection Framework
  • Experimentation Brain Structured Testing Protocol
  • Data Brain Personalization Measurement Framework
  • Customer Brain Segmentation Framework
  • Ecommerce Brain Personalization Journey Framework

Drift Protection

The system must prevent:

  • random personalization
  • untested experiences
  • overuse of rules
  • misuse of predictive systems
  • lack of measurement
  • absence of learning

Architectural Intent

This framework ensures:

  • MWMS experiments intelligently
  • personalization scales effectively
  • insights compound over time
  • decisions are based on data

It transforms testing from:

→ isolated experiments

into:

→ continuous optimization systems

Final Rule

If an experience is not tested:

→ it must not be scaled

Change Log

Version: v1.0
Date: 2026-05-03
Author: HeadOffice

Change:
Created Personalization Testing Framework defining A/B testing, rules-based personalization, predictive personalization, and unified testing workflow based on CXL systems.

Change Impact Declaration

Pages Created:
Experimentation Brain Personalization Testing Framework

Pages Updated:
None

Pages Deprecated:
None

Registries Requiring Update:
Experimentation Brain Page Registry

Canon Version Update Required:
Yes

Change Log Entry Required:
Yes

END MWMS EXPERIMENTATION BRAIN PERSONALIZATION TESTING FRAMEWORK v1.0