Morning readout

Wednesday 29 July 2026 · week 31 · ET — data through Tue 28 Jul

Scaling day — next scale Friday

Blended ROAS

2.43×

target 2.50 · floor 2.25

Spend

$7,277

in-app $5,744 · W2A $1,533

Trials started

982

Singular 778 · W2A 204

ARPT

$15.57

per trial · ET day

Cost per trial

$7.41

target $7.22 · floor $8.02

MON scaleTUE restWED scaleTHU restFRI scaleSATSUN

Yesterday in full

How the blended number is built, line by line.

Line	Value	Derivation
Singular in-app trials	778	Paid only — Organic excluded (1,984)
× 1.2 coefficient	934	Singular under-reports app attribution
W2A trials	204	Meta, face value
Trials started (counted)	982	778 + 204 — the real count
Revenue trials	1,138	934 + 204 — revenue only
ARPT	$15.5719	+ monthly subs at actual gross
Paid revenue	$17,715	1,138 × $15.5719
Meta spend	$7,277.05	App $5,744.42 + W2A $1,532.63
Cost per counted trial	$7.41	÷ 982 — target $7.22
Cost per revenue trial	$6.40	÷ 1,138 — target $6.23
ROAS	2.43×	above the floor, under target

Scale up or down

Proven ad sets and creative tests from weeks 28–30. Tick to include a change; current budget is fixed and only the deltas respond. A test marked CUT is below 2.0× on a budget of $50 or less — switched off rather than trimmed. Everything else still bottoms out at −20%.

In-app — Singular × 1.2

Ad set	Spend	Trials	CPT	ROAS vs floor / target	Current	Change3-day basis

W2A — Meta face value

Ad set	Spend	Trials	CPT	ROAS vs floor / target	Current	Change3-day basis

Confirm and send

Untick anything you don't want actioned. New budgets carry the cents marker — .02 scaled up, .01 scaled down — so direction stays readable in Ads Manager.

Send to Note to add optional

Draft Gmail

This week's tests

Week 31 started Monday. One day old — not rated, not flagged. $400/day between them; if still at zero delivery by Friday it becomes a setup question.

Ad set	Budget	Spend to date	Trials	Status

Where the windows disagree

3-day is the gold star. These sit on opposite sides of the floor depending on the window — a sanity check, never an override.

Ad set	Budget	3-day	7-day	Conflict

Ads burning spend

Live ads only — anything paused is filtered out. ≥$20 spend and CPT above $9.00 against a $7.19 target. A watch-out list, not an instruction.

Ad	Campaign / ad set	Spend	Trials	CPT

Flags

Checked every day, including rest days. Current-week ad sets are excluded — a week-31 test with no delivery is new, not broken.

Model. revenue = ARPT × (Singular × 1.2 + Meta W2A) · ROAS = revenue ÷ Meta spend. Singular is the basis for in-app; W2A at face value.
Thresholds. Floor 2.25×, target 2.50×. Above the floor is profitable at $200k and gets scaled. Proven ad sets are never cut — worst case −20%.
Cutting a test. A prior-week TEST ad set below 2.0× on the 3-day window, on $50/day or less, is switched off instead of trimmed — agreed 29 Jul, and it applies to tests only. Current-week tests are never eligible.
One decision window. Only the 3-day window moves a budget.
Cadence. Mon / Wed / Fri; rest Tue / Thu / weekend. Flags daily.
Naming. CT + ISO week + index. Current week is 31.
Two trial counts. Counted (949) is real; revenue trials (1,096) is grossed up for revenue only. Gross-up 1.1549 — compare a CPT only against a target on its own basis.
Guards. ARPT and Meta both on America/New_York. Organic excluded from Singular. Alignment rows filtered. Ad sets keyed on ID, never name.
Open. Android excluded from ARPT. elevate / ascend rates ASSUMED.