Experiment context
Hypothesis. Moving high-frequency terms into separate low-bid campaigns was expected to cut CPC by 30–50% while maintaining enough traffic for lower CPA.
How the test ran. GA4 data was not used because of technical issues; the readout relies on Google Ads data for moved keyword pools.
Comparison design. Pre 08.01–12.02 vs post 13.02–20.03; no traffic split was used.
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 | Date comparison | Clicks | Impr. | Cost | Avg. CPC | CTR | Interactions | Est. first position bid | Est. first page bid | Est. top of page bid | Quality Score | All conv. | Landing page exp. (hist.) | Impr. (Top) % | Label | Conversions | Conv. rate | Conv. value | Conv. value / cost | Cost / conv. | Original conv. value |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| us nontest | 8 jan - 12 feb | 1,536 | 7,506 | 6,363.42$ | 4.14$ | 20.46% | 1,536 | 0 | 0 | 0 | 142 | 591.06 | 0 | 13.55 | 0 | 129.04 | 8.40% | 8,943.58$ | 1.41 | 49.31$ | 8,943.58$ |
| us test | 13 feb - 20 mar | 2,693 | 17,280 | 4,586.58$ | 1.70$ | 15.58% | 2,693 | 0 | 0 | 0 | 145 | 1,455.15 | 0 | 14.21 | 0 | 328.84 | 12.21% | 28,032.52$ | 6.11 | 13.95$ | 28,032.52$ |
| totals | test/nontest | 75.33% | 130.22% | -27.92% | -58.89% | -23.84% | 75.33% | N/A | N/A | N/A | 2.11% | 146.19% | N/A | 4.87% | N/A | 154.84% | 45.35% | 213.44% | 334.86% | -71.72% | 213.44% |
| Campaign | Date comparison | Clicks | Impr. | Cost | Avg. CPC | CTR | Interactions | Est. first position bid | Est. first page bid | Est. top of page bid | Quality Score | All conv. | Landing page exp. (hist.) | Impr. (Top) % | Label | Conversions | Conv. rate | Conv. value | Conv. value / cost | Cost / conv. | Original conv. value |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| eu nontest | 8 jan - 12 feb | 5,640 | 21,758 | 13,789.30$ | 2.44$ | 25.92% | 5,640 | 0 | 0 | 0 | 243 | 1,399.54 | 0 | 25.16 | 0 | 289.30 | 5.13% | 13,198.43$ | 0.96 | 47.66$ | 13,198.43$ |
| eu test | 13 feb - 20 mar | 6,259 | 30,477 | 8,993.38$ | 1.44$ | 20.54% | 6,259 | 0 | 0 | 0 | 249 | 2,004.34 | 0 | 26.65 | 0 | 456.63 | 7.30% | 32,542.22$ | 3.62 | 19.70$ | 32,542.22$ |
| totals | test/nontest | 10.98% | 40.07% | -34.78% | -41.23% | -20.77% | 10.98% | N/A | N/A | N/A | 2.47% | 43.21% | N/A | 5.90% | N/A | 57.84% | 42.23% | 146.56% | 278.05% | -58.68% | 146.56% |
Conclusions
Moving high-volume keywords into dedicated low-bid campaigns reduced the cost of buying traffic while preserving useful coverage. The structure is worth retaining where volume is sufficient, with bids and negatives reviewed regularly to prevent overlap with the original campaigns.
Limitations. This is a before/after comparison without a concurrent control. Seasonality, demand and auction changes may affect the observed uplift, and GA4 data was unavailable.