The methodology
The Growth Infrastructure Method.
Three years of testing across three businesses. Tool-agnostic by design. Built on principles that survive the next tool cycle.
Access to AI is not the premium. The tools are free or close to it. Judgement at the architectural level is the only thing left that commands a price.
Five design principles
- 01Own the data layer, not the vendor
- 02Voice-first, tool-second
- 03Modular, not monolithic
- 04Human judgement at the edges
- 05Thirty days to live
The shift
AI moved the hard part.
AI has turned the build of a growth system from a six-figure agency project into a thirty-day architectural decision. That is the shift.
The shift has not removed the hard part. It has moved it. The hard part is no longer can we build it. The hard part is what to build, what to reject, and how to make it survive the next tool cycle.
Most solo professionals lose eighteen to twenty-four months to that problem and still have nothing scaling. Not because they are not smart. Because the problem is not a tool problem. It is a judgement problem.
Access to AI is not the premium. The tools are free or close to it. Judgement at the architectural level is the only thing left that commands a price. That means knowing what to build, what to rule out before the build starts, and how to make the result survive when the tools underneath it change.
The five design principles
Every IGP build passes through these five.
If a proposed component breaks any of them, it does not ship.
Own the data layer, not the vendor
Client data lives in systems the client controls. Domain, hosting, ad account, email list, content library — all in your name. IGP is the operator, not the landlord. When a tool dies or triples its price, the data moves. The business does not rebuild.
Voice-first, tool-second
Every system that produces client-facing output is trained on the founder's voice before it is wired to any tool. The voice file is the source asset. The tool is the renderer. Swap the generator in a weekend. Rebuilding a voice takes months.
Modular, not monolithic
The stack is a set of components with clean boundaries. The content engine does not depend on the ads account. The scorecard does not depend on the CRM vendor. Monolithic stacks break as a whole. Modular stacks absorb tool changes as local repairs.
Human judgement at the edges, automation at the middle
The middle is repetitive, high-volume, low-stakes. Generate a draft. Score a lead. Schedule a post. Automate it. The edges are irreversible and reputation-bearing. Publish. Send to a client. Sign a contract. A human reviews every edge. No automation has final say on anything the market can see.
Thirty days to live, compounding thereafter
The full stack stands up in thirty days. Not perfect. Live. Every subsequent week adds to a system that has been shipping to the market since day thirty. Voice-trained content improves with approval data. Scorecard conversion improves with traffic. Ad creative improves with testing cycles. The compounding starts on day thirty.
The discarded options
What the method rejects.
Naming the discarded options matters more than listing the kept ones. The value of a methodology is in what it rules out. All five rejected paths solve for which tool. None of them solve for what architecture.
Tool-centric builds
Starting with which tool you like means re-running the entire build the next time that tool changes. The tool is the last decision, not the first.
Agency bolt-ons
Wrapping a chatbot around a WordPress site. Data layer wrong. Voice layer absent. Stack monolithic. Three out of five principles broken before a single client sees it.
DIY with YouTube
Works for technical tinkerers with unlimited time. Does not work for a solo professional whose hours are the practice.
One-off prompt consultants
Sells prompts. Does not build a system. Prompts age out inside a quarter.
Course-plus-community self-build
A paid course, a Slack community, evenings for a year. Works for the small slice of buyers who finish it. Most do not.
Provenance
Three years. Three builds.
The method was not designed in a whiteboard session. It was built, broken, and rebuilt three times across three businesses.
Imperium Negotiation Solutions
Jan's own negotiation consultancy. The original stack. The testing that produced the principles happened here first.
Linda Paige
A second practice, a different vertical, the same founder requirement. Voice-first, client-owned, thirty-day build. The principles held. The tools in Linda's stack are not all the tools in the Imperium stack. The architecture is.
Imperium Growth Partners
Productising the method so a solo professional can buy in and be live in thirty days without learning any of the underlying tools.
Three builds. Three years. That is what the ICP is paying for. Not access to AI. The judgement of what to build with it.
Jan Potgieter · Founder, IGP
Durability
A tool list ages. This does not.
A tool list needs a refresh every twelve months. A methodology built on fundamentals does not.
- ·Own the data layer, not the vendor. Correct in 2010. Correct in 2030.
- ·Voice-first, tool-second. Correct before LLMs existed. Correct when the next generation arrives.
- ·Modular, not monolithic. A rule from the 1970s software world. It has not stopped being true.
- ·Human judgement at the edges. A regulatory reality for the ICP. Also a moat.
- ·Thirty days to live. A discipline, not a technology. Disciplines do not expire.
When a specific tool in the stack dies, IGP swaps it. The client's system keeps running. The method does not change.
Find out if the method fits your practice.
The Growth Readiness Scorecard takes three minutes. You get a personalised report and a clear view of what a Growth Infrastructure build would look like for your practice.
Take the scorecard →P.S. If you have questions about the method before taking the scorecard, email [email protected]. I read every one.