MWMS MCR

Experimentation Brain Forecast Based Optimization Framework

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

Purpose

The Forecast Based Optimization Framework defines how MWMS uses forecasts to guide testing, decision-making, and optimization.

It ensures that:

  • all experiments have defined expectations
  • all results are measured against benchmarks
  • all optimization decisions are structured
  • all actions are based on data, not opinion

Without forecasting, MWMS risks:

  • unclear test outcomes
  • subjective decision-making
  • inconsistent optimization
  • wasted traffic and budget

Core Principle

Optimization is the comparison between Expected Results and Actual Results.

No experiment or optimization may occur without a defined forecast.

Framework Structure

The framework consists of three layers:

  1. Forecast Definition
  2. Measurement & Comparison
  3. Optimization Decision

1. Forecast Definition

Definition

A forecast defines expected performance before testing begins.

Requirements

Each forecast must include:

  • expected result (end outcome)
  • expected “how” (process performance)
  • acceptable performance range
  • time window

Forecast Types

Result Forecast

Defines final outcome expectations:

  • conversion rate
  • revenue per user
  • cost per acquisition

Process Forecast

Defines expected behaviour within the funnel:

  • click-through rate
  • step conversion rates
  • drop-off points

Example

  • Landing page conversion: 20–30%
  • Add to cart rate: 8–12%
  • Checkout completion: 2–5%

Rules

  • forecasts must be defined before testing
  • forecasts must include ranges (not single values)
  • forecasts must align with system maturity
  • forecasts must be realistic and data-informed where possible

2. Measurement & Comparison

Definition

Actual performance is measured and compared against the forecast.

Components

  • actual results
  • expected range
  • variance (difference between actual and expected)

Types of Outcomes

Within Range

  • performance meets expectations
  • system is functioning correctly

Above Range

  • performance exceeds expectations
  • opportunity to scale

Below Range

  • performance is under expectations
  • indicates problem or inefficiency

Rules

  • always compare to forecast, not raw numbers
  • avoid emotional interpretation
  • focus on variance, not isolated metrics

3. Optimization Decision

Definition

Defines actions taken based on variance between expected and actual results.

Decision Logic

If Above Forecast

  • scale traffic
  • increase budget
  • expand exposure
  • replicate across channels

If Within Forecast

  • maintain performance
  • monitor stability
  • run controlled improvements

If Below Forecast

  • identify weakest stage
  • diagnose cause
  • implement tests or fixes

Optimization Levels

Result Level

  • optimize final outcome (sales, leads, revenue)

Process Level

  • optimize individual steps (funnel stages, user flow)

System Level

  • optimize across platforms (ads, funnel, backend)

Rules

  • always fix weakest link first
  • optimize based on data, not assumptions
  • prioritize high-impact areas
  • do not scale broken systems

Optimization Workflow

  1. Define forecast
  2. Launch test or campaign
  3. Collect data
  4. Compare actual vs expected
  5. Identify variance
  6. Determine weakest stage
  7. Apply optimization action
  8. Re-measure
  9. Repeat

Cross-Brain Integration

Experimentation Brain

Owns:

  • forecast creation
  • variance analysis
  • test design
  • optimization logic

Affiliate Brain

Uses:

  • optimization outputs for scale/kill decisions
  • performance benchmarks for offers

Ads Brain

Uses:

  • traffic-level forecasts
  • optimization for ad performance

Conversion Brain

Uses:

  • funnel-level forecasts
  • optimization for user behaviour

Finance Brain

Uses:

  • forecast vs actual for budget decisions
  • validation before scaling

Failure Conditions

System fails when:

  • no forecast exists
  • results are evaluated without benchmarks
  • optimization is based on opinion
  • actions are inconsistent or unclear
  • weak stages are ignored

Performance Indicators

Optimization effectiveness improves when:

  • forecast accuracy increases
  • variance decreases over time
  • decision speed improves
  • test efficiency improves
  • scaling success rate increases

Relationship to Measurement Matrix

This framework operates across:

  • Forecasting → defines expectations
  • Optimizing → executes actions

It relies on:

  • Planning → defines questions
  • Building → collects data
  • Reporting → provides insights

Outcome

When applied correctly, this framework enables MWMS to:

  • optimize faster
  • reduce wasted spend
  • improve conversion rates
  • scale winning systems with confidence
  • build predictable performance models

End of Framework