Landing page test — project-gamma

Replacing the standard URL with a CTA-style URL was expected to increase commercial relevance, CTR and transactions.

A/B landing-page testSplit: 50/503 performance snapshots
A/Btest design
Split50/50
KPICTR / CR / CPA-CPO / ROAS-CAC

Experiment context

Hypothesis. Replacing the standard URL with a CTA-style URL was expected to increase commercial relevance, CTR and transactions.

How the test ran. The analysis compares regions and landing-page variants across Google Ads and purchase metrics.

Traffic split. 50/50.

Performance results 3 snapshots · Click to expand

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.

EU — project-gamma — original URL vs CTA URL
ClicksImpr.ConversionsCost / conv.conv. rateTransactionsCR (click-transactions)
test1701,7391137.30$6.47%52.94%
original1401,65200$0.00%32.14%

Result. Successful in EU: the CTA URL produced 21% more traffic and 5 Google Ads transactions versus 0 for the original URL.

US — project-gamma — original URL vs CTA URL
ClicksImpr.ConversionsCost / conv.conv. rateTransactionsCR (click-transactions)
test1131,9091.67286.60$1.47%21.77%
original961,7792205.43$2.08%11.04%

Result. Inconclusive in US: the test produced 2 transactions versus 1 for control; retain the original URL until more data is collected.

WW — project-gamma — original URL vs CTA URL
ClicksImpr.ConversionsCost / conv.conv. rateTransactionsCR (click-transactions)
test322751.6758.13$5.21%00.00%
original58334358.04$5.17%46.90%

Result. Unsuccessful in WW: test traffic fell 45% and produced 0 transactions versus 4 for the original URL.

Conclusions

Overall verdict: region-specific. Roll out the CTA URL in EU, where it produced 21% more traffic and 5 Google Ads transactions versus 0 for control. Keep the original URL in WW, where the test lost 45% of traffic and produced 0 transactions versus 4. Treat US as inconclusive because the result was only 2 transactions versus 1.

Limitations. transaction counts are low, especially in US, so small absolute differences are insufficient for a confident conclusion.