Growth Architecture · the methodology
Growth is not a campaign. It is an operating system.
The problem
Eight failure modes that look like marketing problems and are not.
01
Strong marketing, weak positioning
02
High traffic, weak conversion
03
Multiple channels, no coordination
04
Good people, unclear ownership
05
More data, slower decisions
06
Founder-dependent growth
07
Agencies optimising locally
08
AI tools without operating logic
Growth breaks at the joins.
The Growth Architecture · nine layers
Select a layer to see its purpose, its typical failure and the question that tests it.
How the system works
Strategy sets the direction. Measurement feeds optimisation, and optimisation returns to strategy — the system is a loop, not a funnel.
01
Business strategy
Sets what growth is for.
02
Positioning
Defines who demand is built for.
03
Demand
Creates reachable buyers.
04
Conversion
Turns demand into revenue.
05
Customer growth
Compounds what is already won.
06
Data + AI
Makes the system legible and faster.
07
People + execution
Gives it owners and rhythm.
08
Measurement
Reads the constraint honestly.
09
Optimisation
Returns evidence to strategy.
The methodology
Five stages, run in order. Stage one is never skipped, however obvious the answer appears.
Stage 01
Audit
Find the constraint. Read all nine layers together rather than reviewing functions separately.
Leaves behind
Growth Constraint Map
Stage 02
Strategy
Decide where advantage should exist and what the company will deliberately not pursue.
Leaves behind
Strategic direction and trade-offs
Stage 03
Systems
Create ownership, metrics and operating logic for each layer that matters this year.
Leaves behind
Decision rights and metric model
Stage 04
Execution
Run the system on a stated cadence with the team and partners already in place.
Leaves behind
Operating cadence and roadmap
Stage 05
Optimisation
Compound learning: every review either confirms the constraint or moves it.
Leaves behind
Revised constraint and priorities
The constraint map
A worked example. Four signals, read together rather than separately.
Observed signals
Traffic
Up
Leads
Flat
Sales
Inconsistent
CAC
Rising
Read across, not down
Each signal has an obvious local owner and an obvious local fix. Taken together they describe one failure, not four.
Actual constraint
Positioning + conversion handoff
Upstream cause
Positioning stopped filtering. Volume grew because the claim got broader, so a larger share of traffic was never going to buy.
Downstream effect
Sales inherits unqualified demand, follow-up slows, close rate drops, and blended CAC rises even though media efficiency is unchanged.
What not to do first
Do not cut spend. Spend is the symptom’s loudest number and the constraint’s least relevant one.
The joins
Eight joins. Each one is a place where two owners both assume the other has it.
Sales
Both measure their own stage. Neither owns the definition of a qualified lead.
Demand
The claim the market hears is not the claim the campaign was built from.
Conversion
Volume targets are met by loosening exactly what conversion depends on.
Retention
Acquisition is owned by name; retention is owned by whoever notices first.
Decisions
The number exists in a dashboard nobody is required to act on.
Workflow
Output increases, the decision it was meant to improve is unchanged.
Execution
The plan is agreed annually and the priorities are reset weekly.
Team
Decisions are delegated without the standard that would let someone else make them.
Decision rights
Every system in the architecture is defined by the same six fields. A system missing any of them is not a system.
Field 01
Named owner
One person, not a function or a committee.
Field 02
Decision authority
What they can decide alone, and the limit where it changes.
Field 03
Input metric
The leading number they control week to week.
Field 04
Output metric
The result they are judged on over a quarter.
Field 05
Review rhythm
A fixed cadence, with the pack defined in advance.
Field 06
Escalation rule
The condition that puts it on the founder’s desk — and nothing else does.
Worked example · demand layer
Owner: head of demand. Input metric: qualified pipeline created. Output metric: cost per qualified opportunity.
Cadence and escalation
Reviewed weekly against plan, monthly against the constraint. Escalates to the founder only when the output metric misses two consecutive months.
AI inside the architecture
AI is not a tenth layer. Each application is attached to the layer whose decision it changes.
Research
Layer 01
Faster evidence for positioning choices.
Segmentation
Layer 02
Who demand is built for, refreshed continuously.
Forecasting
Layer 02
What to fund next quarter, with confidence bands.
Lead intelligence
Layer 03
Which opportunities get human time first.
Content operations
Layer 02
What gets produced against demand gaps, not calendars.
CRM
Layer 03
When follow-up happens and who owns the next step.
Reporting
Layer 09
What the review reads before the room forms an opinion.
Decision support
Layer 09
Options and trade-offs prepared before the meeting.
Customer service
Layer 04
Which issues signal churn rather than friction.
Knowledge
Layer 07
The standard a new owner decides against.
Automation
Layer 08
Which steps no longer need a person in the loop.
What changes
Before
Channel-led
Reactive
Founder-dependent
Multiple dashboards
Unclear ownership
Agency fragmentation
Tactical AI adoption
After
System-led
Constraint-led
Clear decision rights
Shared metrics
Integrated execution
Documented operating cadence
AI tied to decisions
Where it is used
The engagement changes. The architecture underneath does not.
Advisory
Fractional CMO
The architecture is operated over six to twelve months.
Consulting
Strategic Consulting
One decision, resolved against the same nine layers.
Diagnostic
Growth Audit
The constraint named before anything is bought.
Programme
Growth Sprint
A system designed and installed in ninety days.
Advisory
Founder Advisory
Standing counsel using the architecture as shared language.
Product
Digital Dheeraj
The same layers, available as a strategic interface.
Ventures
Venture building
Where the architecture is applied to businesses I own.
Who it is for
A good fit
Founder-led businesses
Scaleups
Growth-stage companies
Leadership teams
Companies with multiple channels
Teams preparing to scale
Not a fit
Companies looking only for campaign tactics
Teams unwilling to change ownership or operating cadence
Sample diagnostic
Seven questions the architecture asks before it accepts that answer. Select one to see what it is testing.
Before you write
