Experiment context
Hypothesis. Dedicated retention campaigns were expected to lift CTR, CR trans, CR FP and ROAS by separating warm audiences from broad acquisition.
How the test ran. All four project/direction combinations are preserved below with _retention/_all suffixes kept because they identify the audience group.
Comparison design. Retention vs all-traffic baselines across 2 projects × 2 directions.
Performance results
N/A means the metric cannot be calculated because the baseline is zero or the metric does not apply; — means no source value was provided.
| Campaign | Impr. | Clicks | CTR | Cost | Avg. CPC | Transactions | CPO | Total revenue | AOV | ROAS | First purchasers | CR (first purchasers) | CR (transactions) | CAC | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 01.02.26 - 30.06.26 | project-beta_service6_retention | 6,015 | 720 | 12.00% | 1,766.93$ | 2.45$ | 72 | 25.00$ | 6,203.80$ | 86.00$ | 351.00% | 11 | 1.53% | 10.00% | 161.00$ |
| 01.09.25 - 31.01.26 | project-beta_service6_all | 232,250 | 14,983 | 6.50% | 24,112.85$ | 1.61$ | 593 | 41.00$ | 49,129.90$ | 83.00$ | 204.00% | 118 | 0.79% | 3.96% | 204.00$ |
| Campaign | Impr. | Clicks | CTR | Cost | Avg. CPC | Transactions | CPO | Total revenue | AOV | ROAS | First purchasers | CR (first purchasers) | CR (transactions) | CAC | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 01.02.26 - 30.06.26 | project-beta_service3_retention | 13,958 | 4,218 | 30.20% | 9,714.09$ | 2.30$ | 859 | 11.00$ | 47,954.70$ | 56.00$ | 494.00% | 45 | 1.07% | 20.37% | 216.00$ |
| 01.09.25 - 31.01.26 | project-beta_service3_all | 380,314 | 73,453 | 19.30% | 168,683.62$ | 2.30$ | 6,519 | 26.00$ | 316,812.10$ | 49.00$ | 188.00% | 2,405 | 3.27% | 8.88% | 70.00$ |
| Campaign | Impr. | Clicks | CTR | Cost | Avg. CPC | Transactions | CPO | Total revenue | AOV | ROAS | First purchasers | CR (first purchasers) | CR (transactions) | CAC | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 01.03.26 - 30.06.26 | project-gamma_service6_retention | 11,593 | 3,688 | 31.80% | 10,807.26$ | 2.93$ | 523 | 21.00$ | 41,128$ | 79.00$ | 381.00% | 14 | 0.38% | 14.18% | 772.00$ |
| 01.11.25 - 31.01.26 | project-gamma_service6_all | 109,633 | 16,287 | 14.90% | 31,364.88$ | 1.93$ | 1,362 | 23.00$ | 68,884.10$ | 51.00$ | 220.00% | 239 | 1.47% | 8.36% | 131.00$ |
| Campaign | Impr. | Clicks | CTR | Cost | Avg. CPC | Transactions | CPO | Total revenue | AOV | ROAS | First purchasers | CR (first purchasers) | CR (transactions) | CAC | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 01.02.26 - 30.06.26 | project-gamma_service3_retention | 16,064 | 5,508 | 34.30% | 11,934.18$ | 2.17$ | 1,155 | 10.00$ | 71,629.80$ | 62.00$ | 600.00% | 37 | 0.67% | 20.97% | 323.00$ |
| 01.09.25 - 31.01.26 | project-gamma_service3_all | 293,721 | 49,945 | 17.00% | 70,767.02$ | 1.42$ | 6,477 | 11.00$ | 303,868.20$ | 47.00$ | 429.00% | 1,278 | 2.56% | 12.97% | 55.00$ |
Conclusions
Separating retention traffic made the performance of warm audiences easier to control and compare with acquisition. Retention should remain in dedicated campaigns where its stronger intent supports better CR and ROAS, with exclusions used to keep acquisition and retention audiences distinct.
Limitations. Retention and all-traffic campaigns cover different historical periods and audiences with different intent. The comparison demonstrates segment efficiency but does not isolate the causal effect of the campaign split.