Document Type: Framework
Status: Active
Version: v1.0
Authority: MWMS HeadOffice
Parent: Finance Brain Canon
Slug: finance-brain-cohort-revenue-forecasting-framework
Last Reviewed: 2026-04-13
Purpose
The Finance Brain Cohort Revenue Forecasting Framework defines how MWMS models future revenue based on customer cohort behaviour rather than simple top-line averages.
Traditional revenue forecasting often assumes stable averages.
In reality, revenue performance is driven by cohort behaviour patterns.
Different customer groups behave differently over time.
Cohort-based forecasting improves:
• revenue predictability
• growth realism
• retention visibility
• acquisition planning quality
• capital pacing decisions
• survivability awareness
Cohort modelling reveals how revenue accumulates through customer lifecycle behaviour.
Revenue is produced by groups of customers acquired at different points in time.
Understanding how those groups behave improves forecast reliability.
Scope
This framework applies to:
• cohort-based revenue projections
• new customer revenue modelling
• repeat purchase modelling
• cohort decay curve interpretation
• lifecycle revenue contribution analysis
• customer file durability modelling
• revenue accumulation timing interpretation
This framework governs how Finance Brain structures forward-looking revenue projections using customer cohort behaviour.
It does not govern:
• capital approval decisions
• lifecycle campaign execution
• CRO optimisation decisions
• experiment validation authority
• traffic allocation decisions
Those remain governed by Finance Brain capital logic, Ecommerce Brain lifecycle systems, Experimentation Brain, and Ads Brain.
Definition / Rules
Core Principle
Revenue is generated by customers over time.
Customers do not behave as a single uniform group.
Different acquisition periods produce different behavioural patterns.
Cohorts represent groups of customers acquired within the same timeframe.
Each cohort demonstrates behaviour patterns such as:
purchase frequency
repeat purchase timing
spend accumulation
engagement persistence
decay speed
Cohort modelling captures how revenue accumulates through customer lifecycle activity.
Cohort Definition
A cohort is a group of customers acquired during the same period.
Examples:
customers acquired in a specific month
customers acquired through a specific campaign period
customers acquired through a specific channel
Each cohort contributes revenue across multiple future periods.
Revenue impact extends beyond initial purchase.
Cohorts generate revenue curves.
Cohort Revenue Curve
A cohort revenue curve describes how revenue accumulates over time from a group of customers.
Typical curve characteristics:
initial purchase spike
second purchase drop-off
gradual repeat purchase tail
long-term residual value contribution
Understanding curve shape improves forecast realism.
Revenue is rarely linear.
Revenue typically decays gradually over time.
Decay speed influences forecast stability.
New Customer Contribution
Forecast must incorporate expected revenue contribution from newly acquired customers.
Variables influencing contribution:
acquisition volume
conversion rate
average order value
offer positioning
market conditions
New customer contribution determines initial revenue growth rate.
Acquisition performance influences cohort size.
Repeat Purchase Contribution
Repeat purchase behaviour often produces significant share of total revenue.
Variables influencing repeat contribution:
second purchase probability
purchase interval timing
product replenishment logic
brand affinity strength
lifecycle communication effectiveness
Repeat purchase modelling improves forecast stability understanding.
Second purchase behaviour often represents the most sensitive lifecycle event.
Active Customer File Contribution
Revenue is influenced by the size and behaviour of the active customer file.
Active file characteristics:
number of purchasing customers
engagement persistence
purchase recency distribution
behaviour frequency distribution
Active file durability improves forecast resilience.
Weak active file durability increases dependence on acquisition.
Cohort Decay Behaviour
Cohorts lose purchasing activity over time.
Decay behaviour influences revenue stability.
Decay characteristics include:
speed of activity reduction
frequency of repeat purchasing decline
engagement persistence variation
Rapid decay increases acquisition dependence.
Slow decay improves lifetime value durability.
Cohort Layer Structure
Forecast structure should separate:
new customer cohorts
recent cohorts
mature cohorts
Different cohort ages contribute revenue differently.
Understanding age distribution improves revenue expectation accuracy.
Monthly Projection Logic
Monthly revenue projection should consider:
expected new customers
expected repeat purchase behaviour
expected cohort decay patterns
expected AOV variation
Monthly projections improve operational planning clarity.
Weekly Projection Logic
Weekly projections allow faster responsiveness to behavioural changes.
Weekly signals provide:
early detection of acquisition performance changes
early detection of repeat behaviour variation
early detection of lifecycle disruption
Shorter intervals improve reaction speed.
Cohort Behaviour Signals
Cohort modelling provides insight into:
repeat purchase timing patterns
customer value accumulation speed
seasonality effects
promotion response behaviour
lifecycle messaging effectiveness
Behaviour signals improve strategic decision confidence.
Relationship to Forecast Sensitivity Framework
Cohort behaviour assumptions influence sensitivity risk.
Changes in repeat behaviour can materially alter forecast outcomes.
Sensitivity analysis should evaluate:
repeat rate variation impact
decay speed variation impact
AOV variation impact
Cohort structure informs sensitivity interpretation.
Relationship to Percentile Scenario Forecasting Framework
Percentile scenarios provide outcome ranges.
Cohort modelling informs probability distribution realism.
Scenario reliability improves when cohort behaviour is understood.
Relationship to Lifecycle Brain Systems
Lifecycle systems influence repeat behaviour.
Lifecycle improvements alter cohort revenue curves.
Lifecycle experimentation should consider impact on cohort durability.
Improved lifecycle systems improve forecast stability.
Relationship to Acquisition Systems
Acquisition quality influences cohort behaviour.
Different traffic sources may produce different cohort durability patterns.
Traffic quality variation influences revenue stability.
Acquisition optimisation should consider cohort performance, not only initial CPA.
Planning Use
Cohort forecasting should be used in:
growth planning
revenue expectation modelling
acquisition pacing decisions
lifecycle investment decisions
capital allocation planning
Cohort awareness improves strategic confidence.
Failure Modes Prevented
This framework prevents:
assuming all customers behave identically
over-reliance on simple revenue averages
ignoring repeat purchase contribution
underestimating impact of cohort decay
assuming stable lifecycle behaviour
ignoring differences between acquisition periods
Cohort modelling improves forecast realism.
Drift Protection
The system must prevent:
using blended averages without cohort visibility
ignoring lifecycle influence on revenue durability
assuming repeat purchase behaviour remains constant
treating early cohort performance as permanent behaviour
ignoring cohort decay acceleration signals
Cohort modelling must remain behaviour-based.
Architectural Intent
Finance Brain Cohort Revenue Forecasting Framework exists to ensure MWMS forecasts reflect real customer behaviour patterns rather than simplified averages.
Customer groups behave differently over time.
Understanding cohort behaviour improves forecast accuracy, growth pacing discipline, and capital risk visibility.
Behaviour-based forecasting improves survivability confidence.
Change Log
Version: v1.0
Date: 2026-04-13
Author: MWMS HeadOffice
Change:
Initial creation of cohort-based forecasting framework defining revenue curve modelling, repeat purchase contribution logic, decay behaviour interpretation, and lifecycle-linked forecast structure.
END – FINANCE BRAIN COHORT REVENUE FORECASTING FRAMEWORK v1.0