Experiment context
Hypothesis. Higher bids were expected to buy more impression share, improve CTR and increase conversions while lowering CAC.
How the test ran. The readout combines Google Ads and GA4 purchase metrics after excluding the learning period.
Traffic split. 50/50.
Visual summary — test-bids uplift
The higher-bid variant improved the auction visibility and new-customer efficiency metrics targeted by the hypothesis.
First-purchase CAC−31.99%62.58$ → 42.56$
Search impr. share+22.20%47.12% → 57.58%
Search lost IS (rank)−19.78%52.88% → 42.42%
Performance results
| Campaign | Impr. | Clicks | Cost | CTR | Avg. CPC | Search impr. share | Impr. (Abs. Top) % | Impr. (Top) % | Search lost IS (rank) | Conversions | CPA | Conv. value | Search top IS | Conv. rate | Search lost IS (budget) | Session campaign | Transactions (all users) | Total revenue (all users) | CPO | AOV | Transactions (first purchasers) | Total revenue (first purchasers) | CAC | FP AOV |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| google_search_us_game-2_service3_c1 | 1,489 | 220 | 500.65$ | 14.78% | 2.28$ | 47.12% | 28.45% | 93.37% | 52.88% | 29.16 | 17.17$ | 1,122.56$ | 44.08% | 13.26% | 0.00% | google_search_us_game-2_service3_c1 | 21 | 848.30$ | 23.84$ | 40.40$ | 8 | 327.93$ | 62.58$ | 40.99$ |
| google_search_us_game-2_service3_c1 test bids | 1,278 | 221 | 595.89$ | 17.29% | 2.70$ | 57.58% | 22.15% | 95.18% | 42.42% | 48.58 | 12.27$ | 1,670.11$ | 54.82% | 21.98% | 0.00% | google_search_us_game-2_service3_c1 test bids | 21 | 843.57$ | 28.38$ | 40.17$ | 14 | 363.20$ | 42.56$ | 25.94$ |
| change, % | -14.17% | 0.45% | 19.02% | 16.98% | 18.42% | 22.20% | -22.14% | 1.94% | -19.78% | 66.60% | -28.54% | 48.78% | 24.36% | 65.76% | N/A | — | 0.00% | -0.56% | 19.02% | -0.56% | 75.00% | 10.76% | -31.99% | -36.71% |
Conclusions
The test campaign gained impression share, CTR and conversion rate, while first-purchase CAC improved despite higher CPC. The result supports higher bids for this campaign group, with continued monitoring to confirm that the efficiency gain persists beyond the test window.