Interactive walkthrough · synthetic data · nothing computes or sends

The Growth Experiment Engine

How I run growth: as an instrumented experiment queue, not a channel wishlist. Built here for the same fictional client as the GTM machine, Calyx Diagnostics. Agents watch the funnel and propose tests; humans decide what ships. Every decision has a number attached.

AGENT, measures & proposes HUMAN GATE, ship/kill decisions KILLED, negative or flat readout

All numbers synthetic. The discipline is real: hypothesis → test → readout → decision, nothing ships on vibes.

Funnel, this quarter

source: site analytics + crm + billing, refreshed nightly by agent

Site visitors

18,420

→ 4.1% to lead

Qualified leads

755

→ 22% to demo

Demos held

166

→ 19% to close · watch: down from 26%

Closed

$412K new ARR

Experiment queue

agents propose from funnel signals · humans approve, ship, or kill

running · week 3 of 6

EXP-07, Lead form: clinical proof upfront

split: 50/50 · primary: visitor→lead · guardrail: lead quality score

Variant leads with the validation study ("30% fewer hospitalizations" analog for Calyx: lab-accuracy stat) above the fold. Interim: +0.6pts, not yet significant. No peeking decision until week 6.

human gate · readout ready

EXP-06, Lifecycle: onboarding nudge sequence

ran 6 weeks · n=214 new accounts · primary: first-order completion

+11ptsfirst-order completion −3.2 daystime to first order flatsupport tickets (guardrail held)

Agent recommendation: ship to 100%, fold into onboarding playbook, retire manual check-ins it replaces.

✓ shipped to 100%, playbook updated, owner assigned, revisit in Q3

✗ killed, archived with readout attached

killed · week 6 readout

EXP-05, Paid retargeting: abandoned demo bookings

ran 6 weeks · primary: recovered bookings · CAC guardrail

+0.4ptsrecovery (n.s.) 2.7×target CAC

Flat lift at unacceptable CAC. Killed at readout, budget reallocated to EXP-06 rollout. Kills are outcomes too, the readout is in the library.

ICP tiers

agent-maintained from closed-won patterns · reviewed monthly

Segment	Tier	Why
DTC wellness brands, $5–50M rev	Tier 1	Fastest close (31d), highest retention; kit volume fits ops
Specialty clinics, multi-site	Tier 1	Largest ACV; needs clinical-workflow selling motion
Supplement startups, <$5M	Tier 2	High volume, small orders; route to self-serve tier
Enterprise retail wellness	Tier 3	18-mo cycles; pursue only with warm exec intro

agent · weekly digest to #growth

What moved this week

Demo→close still soft (19%). EXP-06 shipped, +11pts onboarding completion. EXP-07 mid-flight, no decision. One proposal awaiting review. Tier-1 pipeline coverage: 3.1× target.