Build, maintain, optimise, manage: your first 90 days
The four verbs of a growth machine are build, maintain, optimise and manage, and they map cleanly onto a first ninety days. This is the sequence that takes you from a page with four chambers drawn on it to a machine that runs itself with you on the joints.
Build: the spine and the four chambers, once, end to end
The first job is to build the whole thing, not a phase-one slice you promise to finish later. A half-built machine is just a funnel with extra leaks, because the chambers only compound when they are all wired together on one spine.
- Stand up the spine first, the wired data layer every agent will read, so chambers can talk to each other from day one.
- Wire all four chambers end to end, even if each starts simple: demand capture, sixty-second activation, a routing pipeline, a renewal play.
- Resist the urge to add tools. One connected spine beats nine disconnected ones, and every point solution you avoid is a leak you never have to plug.
Build complete, then improve. A machine that is whole but modest beats a machine that is sophisticated in one chamber and absent in the next three.
Maintain: instrument the gauges from day one
The moment the machine is wired, instrument it, before you optimise anything. You cannot fix a joint you cannot see.
- Instrument speed-to-lead first, because the activation joint is the cheapest, highest-ROI fix in the machine and the one that leaks most by default. Know your response time in minutes, not hours.
- Stand up the gauge board, CAC, LTV, the LTV:CAC ratio against the 3-to-1 floor, and CAC payback against its per-segment line.
- Measure pipeline and revenue, not lead count. Wire the bottom-of-machine truth, not the top-of-machine vanity.
Optimise: find the steepest drop, fix it, re-measure
With gauges live, optimisation is no longer guesswork, it is theory of constraints. Walk the joints, find the steepest drop against benchmark, and fix that one thing.
Resist optimising evenly across the machine. Because the stages multiply, a 20 percent lift at the single weakest joint beats a 20 percent lift spread thin across all of them, and it certainly beats doubling top-of-funnel spend into the same holes. Fix the worst joint, re-measure to confirm it held, then move to the next worst. One joint at a time, proven each time.
Manage: mind the joints, run the volume with agents
The steady state is the founder minding the joints while agents run the chambers. Your job is no longer to do the volume work, it is to watch the gauge board, catch the joint that starts to drop, and own the trust-bearing moments, positioning, the offer, the close, where the founder's executive-level trust moves the number an early rep cannot.
From here, each chamber has its own depth, and each rewards its own focused playbook: tracking the flywheel that ties the gauges together, AI-run lead generation for the demand chamber, lead qualification as a filter for the pipeline joint, converting more pipeline through the close, customer retention for the compounding fourth chamber, and the go-to-market strategy that points the whole machine. Build the machine once, instrument it honestly, fix the worst joint first, and then do the thing the old model never let a founder do, step back and run a full-funnel growth engine with far less cost and your hands on only the joints that matter.