Paid acquisition: proven, improving fast — paused by choice.

Eighteen days of Apple Search Ads across the US and Anglo storefronts, run as two creative iterations. The second run's new store images transformed the funnel; the campaign was paused at its efficiency peak pending resolution of store-labeling rights questions — not for performance.

$2,029

total spend

713

installs (Apple-attributed)

$2.85

blended CPA

$1.41

CPA, final 3 days

5.1× better than run 1

The two runs

New store images: tap-through ×10, cost per install ÷3.4.

Run	Dates	Spend	Installs	Tap-through	CPA
Run 1 · original creative	Jun 30 – Jul 6	$712	99	0.44%	$7.19
Run 2 · new store images	Jul 9 – Jul 17	$1,317	614	4.27%	$2.14
Run 2 · final 3 days	Jul 15 – Jul 17	$346	246	8.9%	$1.41

Two levers moved it. Run 2 changed the store images AND dropped manual bidding. Like-for-like (Maximize Conversions only), the new images cut CPA from $3.23 to $2.14 and lifted tap-through from 3.1% to 4.3% — peaking at 8–10% in the final days, when installs ran at $1.41 and the last day at $1.19.

By bid strategy

Manual bidding cost 6.8× more per install.

Strategy	Spend	Installs	Tap-through	CPA
Manage Bids (manual)	$502	33	0.31%	$15.22
Maximize Conversions	$1,527	679	4.11%	$2.25

Manual bidding ran only in run 1 and was killed with it — it bought broad, low-intent impressions (79K impressions, 0.31% tap-through). From Jul 9 every dollar went through Apple's Maximize Conversions, at $2.25 per install overall and $2.14 in run 2.

By storefront

The US was the market. The rest were probes.

Storefront	Spend	Installs	Tap-through	CPA
United States	$1,894	693	1.80%	$2.73
South Africa	$69	9	0.22%	$7.64
Canada	$38	8	0.85%	$4.71
Australia	$21	1	0.32%	$21.24
United Kingdom	$7	1	0.21%	$6.98

93% of spend and 97% of installs were the US storefront at $2.73 — the only market that got real budget. The other four were probe-sized ($7–69 total each): directionally US-like in Canada, weak impression quality in South Africa, and too small to conclude anywhere. Non-US efficiency is untested, not disproven.

Daily cost per install

Still getting cheaper when it stopped — on ~$127 a day.

Bars = daily spend (avg $127/day while active, peak $280 on Jul 11); line = daily CPA. Week-over-week CPA: $7.19 → $2.79 → $1.51, still falling on the final day. Jul 7–8 = the between-runs pause; the Jul 13 bump was a one-day broad-reach test (impressions ×5), corrected the next day. Total outlay: $2,029 over 16 active days.

Why it stopped The campaign was paused on Jul 17 at its efficiency peak — not for performance. Store-labeling rights questions arose that needed resolving before continuing paid distribution, and spend was voluntarily halted while they are worked through. The playbook — audience, keywords, and the winning creative — is retained and restartable.

What it proved 1. Paid works: $1.41–2.14 CPA in the US at small budgets, with 97% of installs from the US storefront ($2.73 all-run US CPA). 2. Creative iteration pays: one image refresh moved every funnel metric by 3–10×. 3. Quality held: paid installs passed the first-session quality gate at the same ~90% rate as organic, and the install base kept its organic floor after the stop — paid was a dial on top of organic demand, not a substitute for it.

Source: Apple Search Ads console performance export (report generated Aug 4, 2026) · CPA = spend ÷ Apple-attributed installs · tap-through = taps ÷ impressions · countries: US, CA, UK, AU, ZA