Insights
How far to trust the numbers, revenue against spend, and the store page.
How platforms count, and what Loop compares
Cross-platform comparison: Each platform’s own count — not a common basis
Platforms count conversions differently (windows, view-through, modeling), so this comparison favors the most generous counter. Loop lowers its confidence and caps budget moves.
| Platform | Counts conversions | Click-through split | Latest lift test |
|---|---|---|---|
| Meta Ads | Not recorded yet | No | None — cross-platform moves are capped |
| Apple Ads | Not recorded yet | No | 14% incremental (6–22%) · switch-off · Sep 13 |
| Google Ads | Not recorded yet | No | None — cross-platform moves are capped |
| TikTok Ads | Not recorded yet | No | None — cross-platform moves are capped |
Ad spend (7d)
$8,581
Installs
4,315
The ad platform's own reported count · iOS and Android
CPI (blended)
$2.73
all spend / 3,141 paid installs (organic excluded)
Ad spend and installs (daily)
How spend and acquisition move together. Divergence is an early sign that efficiency is changing.
Ad spend
$1,292$0
Installs
6580
Oct 11Oct 17
How honest this measurement is
We do not assert more than we know: probabilistic attribution and caveats are stated explicitly.
20% of iOS installs are attributed probabilistically through SKAN/ATT (per-user attribution is broken). CPI does not need per-user attribution — but that same modelled count is its divisor, so CPI is an estimate too. Only spend is exact. How installs are counted →
The install denominator is the platform's own number — nothing checks it
CPI = spend / paid installs: 3,141 of the 4,315 installs counted by the ad platform's own reported count. Installs that no ad spend bought are left out of the divisor. This workspace holds no store-side install count for this window, so nothing checks this number.
To check the divisor, import the store's own install counts (App Store Connect first-time downloads, Play Console installs) as an installs file with authority store_authoritative
Purchase counts are as reported by the channel (no reconciliation configured)
All 157 purchases in this window are channel-reported. Channels tend to over-count their own results, so cost per purchase is not a true value until it is reconciled against your own measurement (store or payment side). Set mediaPurchaseDiscount (a reconciliation factor) in the policy and we will show the reconciled cost per purchase alongside it.
Not configured, so no reconciliation was applied (we are not assuming 1.0)