Experiment context
Hypothesis. Generic service/category semantics were expected to reach new non-game-specific audiences and generate additional transactions at acceptable ROAS/CPO/CAC.
How the test ran. The test was run in stages because initial efficiency was weak; query cleanup and exact-match refinement were part of the process.
Traffic split. 50/50.
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.
| Period | Spend | Conversions (Google Ads) | Transactions | CPO (GA4) | First purchasers | CAC (GA4) |
|---|---|---|---|---|---|---|
| Before optimization (22.05 – 03.06) | 5,438.65$ | 44.46 | 32 | 169.96 | 4 | 1,359.66 |
| After optimization (03.06 – 09.06) | 321.00$ | 10 | 18 | 17.83 | 2 | 160.50 |
Conclusions
Generic service semantics generated incremental reach beyond game-specific queries, but performance should be judged against the same ROAS, CPO and CAC thresholds as core acquisition. Keep the structure isolated so budgets, negatives and bids can be controlled without distorting game-specific campaigns.
Limitations. The before and after periods differ in duration and the campaign setup was optimized between periods. Treat the result as directional rather than as a controlled A/B estimate.