What the reports show
USD 2.30 → USD 2.05 · 393 → 348 reported conversions
The review covers one Philippines Push campaign and one Interactive Ads campaign, anonymised for publication. It compares 8–22 September with 23 September–7 October 2026. These are observations from running campaigns, not a randomised test or a claim about all account activity.
Spend, volume and CPA
| Campaign | Period | Spend (USD) | Reported conversions | CPA (USD) |
|---|---|---|---|---|
| Push | 8–22 Sep 2026 | 903.481 | 393 | 2.299 |
| Push | 23 Sep–7 Oct 2026 | 713.359 | 348 | 2.050 |
| Interactive | 8–22 Sep 2026 | 212.259 | 46 | 4.614 |
| Interactive | 23 Sep–7 Oct 2026 | 129.057 | 40 | 3.226 |
Push CPA declined 10.8%; Interactive CPA declined 30.1%. Conversion volume also fell: 393 to 348 for Push, and 46 to 40 for Interactive. Efficiency and volume should be considered together.
CPA = spend ÷ reported conversions. Changes use unrounded values. Spend is shown to three decimals to preserve the daily-report totals; small differences from whole-period or hourly totals can result from rounding.
Cross-checking with RedTrack
The destination tracking link was matched to the RedTrack campaign, then separated by the originating ad campaign ID. For the same Push campaign, RedTrack recorded 895 → 781 default conversion events and a 6.5% decline in cost per default event. The Interactive campaign recorded 100 → 86 default events and a 27.0% decline. Both sources point in the same direction, but their counts and costs do not reconcile.
RedTrack event definitions, currency, deduplication and timezone still need verification. The reports are therefore shown separately. A platform conversion is not yet a verified unique sign-up, first-time deposit (FTD), paying customer or profit.
Time-of-day findings
Push performance was grouped into six-hour blocks. All times below are PropellerAds reporting time (EST), not Philippines local time.
| Time block (EST) | Earlier CPA / events | Later CPA / events |
|---|---|---|
| 00:00–05:59 | 3.880 / 23 | 4.632 / 14 |
| 06:00–11:59 | 2.030 / 214 | 1.800 / 221 |
| 12:00–17:59 | 1.478 / 83 | 1.785 / 66 |
| 18:00–23:59 | 3.523 / 73 | 2.830 / 47 |
06:00–17:59 was the stronger combined window in both periods: CPA 1.876 with 297 events, then 1.796 with 287 events. Other hours averaged 3.608 with 96 events and 3.243 with 61 events. This is a candidate for testing, not an automatic instruction to increase bids.
Hourly RedTrack data was unavailable in this review. Interactive Ads had only 6–21 events per six-hour block in each period, so finer dayparting would overstate the evidence. Uneven delivery, traffic mix and conversion delay may also affect hourly results.
The next optimisation test
- Confirm the event definition, report and campaign timezones, deduplication, attribution date basis and conversion delay.
- Run a pre-defined 14-day Push schedule test after reconciliation, keeping creative, geography and total budget stable and preserving a comparison group.
- Judge valid conversion cost together with volume and downstream quality; wait for delayed outcomes before deciding. A fixed number of days alone does not guarantee enough evidence.
This is the proposed next test. The review did not change campaign schedules, bids or budgets.
How to read this case
The two windows have equal duration but different weekday composition. The review does not isolate which previous campaign change caused the improvement, establish statistical significance or verify revenue. It is separate from the historical deposit CPA 2.27 → 1.72 example (currency unconfirmed) elsewhere on this website. Results are specific to these campaigns and periods.
Data and methodology
Source: authenticated, read-only PropellerAds and RedTrack report queries on 8 October 2026. The main table uses summed daily PropellerAds reports; timezone defaults to EST. The latest reporting day may still receive late conversion postbacks. Campaign names, IDs, tracking URLs and customer details are omitted.
Report downloads are not shared through this portfolio.