008 · Experimentation
The standing rule made real: measured changes ship against controls — Growthbook infra live, a weekly cadence, and a dedicated experimentation contractor from November
What this is
Experimentation as the layer under everything: the H2 plan's standing rule — measured changes ship against controls, and contested decisions get settled with data. The infrastructure already exists (Growthbook wired through Next.js middleware, with duplicate-any-page A/B capability — WM built it in April); this project turns infrastructure into a program: an operating doc, a weekly cadence, and a dedicated Growth Specialist, Experimentation contractor once agency contracts end in November. First ship (memo): operating doc + weekly cadence.
Goal: testing as infrastructure — every landing page ships A/B-by-default, and the roadmap's contested calls get decided by lift, not seniority.
Why now
The H2 plan runs on measured lift: treated destination pages against locked baselines, the first user-flow optimization behind a test, paid's two-variable testing system, the activation journey shipped behind a test. All of that assumes a working experimentation rail with an owner — and today the rail exists but the program doesn't. The GA4 sign-off (Data & tracking) makes results trustworthy at exactly the right moment.
The program, per piece
- The operating doc — what qualifies as a test, how tests pre-register, minimum run lengths, who calls results, and the two-variable-per-test / judged-by-concept rule from the paid system generalized.
- The weekly cadence — a standing review: what launched, what read out, what ships next; program memory so learning compounds instead of evaporating.
- The infrastructure, owned — Growthbook + the Next.js middleware rail (any editor duplicates a page, links it to an experiment, and the test runs) maintained in-house post-WM; the LP testing strategy (web workflow, Month 2) runs on it.
- The contractor (November) — the Growth Specialist, Experimentation: plan-design-build-test, the fractional owner who keeps velocity when the founding energy fades.
- The portfolio — tests spanning web, landing pages, email (the activation test), and paid (the format experiments), all reading from the same GA4/warehouse truth.
Work to be done
- Operating doc + weekly cadence live — the rules published, the review running, tests pre-registering — the first ship.
- The rail owned in-house — Growthbook infrastructure transferred from WM knowledge to documented ownership; LP testing running through it from Month 2.
- The contractor onboarded and the portfolio pacing — the experimentation specialist in seat (Nov), a standing test portfolio across web/LP/email/paid with results feeding the weekly attainment story.
Current state
Standing up as the transitions complete: the infrastructure is live (WM-built), the operating doc + cadence is the first ship, and the contractor line opens in November. The website bet's baseline-lock discipline is already the program's method applied.
Dependencies & risks
WM's infrastructure knowledge must transfer before Oct 31 (rides the Storybook ownership doc). Result trust depends on the GA4 sign-off — no sign-off, no believable lift. Until the contractor lands, cadence ownership sits with Scott's pod — keep the weekly light or it won't survive the gap. [owner to complete]
What this is
Experimentation as the layer under everything: the H2 plan's standing rule — measured changes ship against controls, and contested decisions get settled with data. The infrastructure already exists (Growthbook wired through Next.js middleware, with duplicate-any-page A/B capability — WM built it in April); this project turns infrastructure into a program: an operating doc, a weekly cadence, and a dedicated Growth Specialist, Experimentation contractor once agency contracts end in November. First ship (memo): operating doc + weekly cadence.
Goal: testing as infrastructure — every landing page ships A/B-by-default, and the roadmap's contested calls get decided by lift, not seniority.
Why now
The H2 plan runs on measured lift: treated destination pages against locked baselines, the first user-flow optimization behind a test, paid's two-variable testing system, the activation journey shipped behind a test. All of that assumes a working experimentation rail with an owner — and today the rail exists but the program doesn't. The GA4 sign-off (Data & tracking) makes results trustworthy at exactly the right moment.
The program, per piece
- The operating doc — what qualifies as a test, how tests pre-register, minimum run lengths, who calls results, and the two-variable-per-test / judged-by-concept rule from the paid system generalized.
- The weekly cadence — a standing review: what launched, what read out, what ships next; program memory so learning compounds instead of evaporating.
- The infrastructure, owned — Growthbook + the Next.js middleware rail (any editor duplicates a page, links it to an experiment, and the test runs) maintained in-house post-WM; the LP testing strategy (web workflow, Month 2) runs on it.
- The contractor (November) — the Growth Specialist, Experimentation: plan-design-build-test, the fractional owner who keeps velocity when the founding energy fades.
- The portfolio — tests spanning web, landing pages, email (the activation test), and paid (the format experiments), all reading from the same GA4/warehouse truth.
Work to be done
- Operating doc + weekly cadence live — the rules published, the review running, tests pre-registering — the first ship.
- The rail owned in-house — Growthbook infrastructure transferred from WM knowledge to documented ownership; LP testing running through it from Month 2.
- The contractor onboarded and the portfolio pacing — the experimentation specialist in seat (Nov), a standing test portfolio across web/LP/email/paid with results feeding the weekly attainment story.
Current state
Standing up as the transitions complete: the infrastructure is live (WM-built), the operating doc + cadence is the first ship, and the contractor line opens in November. The website bet's baseline-lock discipline is already the program's method applied.
Dependencies & risks
WM's infrastructure knowledge must transfer before Oct 31 (rides the Storybook ownership doc). Result trust depends on the GA4 sign-off — no sign-off, no believable lift. Until the contractor lands, cadence ownership sits with Scott's pod — keep the weekly light or it won't survive the gap.