Autopilot · growth · Sept 2026

The first week.

Two experiments designed so neither can be blamed on the other, and a decision written before the data arrives.

Autopilot app
Portfolio screen
Performance screen
Brokerage select
Danilo Vicioso — Growth · v1.2
Contents Week one · eight parts
Why not just ship both

A new page and new ads, launched together, tell you nothing.

Two variables moved at once. Whatever the number does, both sides of the room can claim it and neither can prove it. So I cross them instead — each arm holds one side constant.

Your page — the control destination
Your current landing page
My page — the variable
The page I built
Arm A — your ads, my page delta = the page
A Autopilot Sponsored

They know something we don’t. Now you can see it.

joinautopilot.com Follow the Pelosi Tracker Install
Fig. 1 — your live creative
→
joinautopilot.com/lp/v2
The landing page I built
{{ n.i }} {{ n.t }}
Fig. 2 — my page, on your domain
Arm B — my ads, your page delta = the creative
A Autopilot Sponsored

Congress has 45 days to tell you what they bought. Day 46, you buy it too.

joinautopilot.com The 45-day rule, automated Install
Fig. 3 — my creative
→
joinautopilot.com
Your current landing page
{{ n.i }} {{ n.t }}
Fig. 4 — your page, untouched

Four cells. Two of them run this week.

{{ c.tag }} {{ c.title }} {{ c.flow }} {{ c.note }}
The read is sized before it starts

Whatever Monday’s baseline turns out to be, I size the read off it before spending.

At a 6% landing conversion it takes about 3,100 sessions an arm to detect a 25% relative lift at 80% power, α = 0.05 — roughly five days at $460/day each. The real figure is computed Monday from your own baseline. Below whatever it comes out to, I don’t call the test — I say what it needs and keep it running.

Monday to Wednesday.

{{ d.day }} {{ d.tag }}
{{ r.t }} {{ r.w }}

Thursday to Sunday.

{{ d.day }} {{ d.tag }}
{{ r.t }} {{ r.w }}

The account, as it will actually be named.

Convention is BRAND_WEEK_ARM_variables_version. Identical targeting on both arms, budgets locked equal.

AP_WK1_ARMA_yourads_mypage_v1 $460/day ├─ ADSET_US_25-54_broad_advplus cloned from your top adset │ ├─ AD_yours_pelositracker_static_01 → /lp/v2?utm_content=a01 │ └─ AD_yours_inversecramer_video_02 → /lp/v2?utm_content=a02 AP_WK1_ARMB_myads_yourpage_v1 $460/day ├─ ADSET_US_25-54_broad_advplus identical, no exclusions │ ├─ AD_dv_45dayrule_static_01 → /landing?utm_content=b01 │ └─ AD_dv_congressfeed_ugc_02 → /landing?utm_content=b02 AP_CONTROL_yourads_yourpage live · untouched all week
Fig. 5 — campaign structure, two arms and one control

The tracking plan, before the first dollar.

1 of 2
Event Parameters Fires Status
{{ e.event }} {{ e.params }} {{ e.fires }} {{ e.status }}

Down the funnel, into the app.

2 of 2
Event Parameters Fires Status
{{ e.event }} {{ e.params }} {{ e.fires }} {{ e.status }}
Fig. 6 — event schema, 24 rows, three of them new

Every outcome has a move already written.

Pre-registered 3 Sept 2026 · v1.2 before any spend
A {{ o.a }} B {{ o.b }}
{{ o.means }} {{ o.move }}

Every morning, in #growth.

format sample — the shape, not the results
Yesterday
{{ d.l }} {{ d.v }}
{{ s.k }} {{ s.t }}

Friday, in writing.

format sample — the shape, not the results
{{ n.l }} {{ n.v }}
{{ s.k }} {{ s.t }}

Then it iterates, every week, the same way.

{{ s.n }} {{ s.title }} {{ s.body }}

And what gets killed, by name.

Concepts get one week, not three — and the reason is written next to each so the same idea doesn’t come back in a different font. These are my own concepts, named; the CTRs get filled in from the real week.

{{ k.name }} {{ k.ctr }}
{{ k.why }}
Fig. 7 — the cut list format · CTR and names filled from the real week

The system behind it.

One number to protect, seven named ways it breaks, and a capability facing every one of them.

Autopilot
Danilo Vicioso · danilojvicioso@gmail.comWeek one · v1.2 · revised 4 Sept 2026