Analysis of 1.82M visitor sessions evaluating layout, copy, and primary call-to-action performance across the full purchase funnel.
Variant F improved every single stage of the funnel over Control A: more visitors clicked (+8.2% relative), more clickers added to cart (+5.2%), and more cart-adders bought (+5.0%). Those compounding gains produced a +19.3% surge in purchases.
If we only measured CTA clicks, C would look like a triumph (48.9% CTR vs. A's 35.0%). However, C's cart-to-purchase rate collapsed to 24.9% (vs. A's 35.1%). It drove low-intent curiosity clicks that abandoned once they saw product details/pricing.
Variant E had the lowest click rate of all versions (30.5% CTR), yet visitors who did click converted at the highest rate in the test (40.1% cart-to-purchase). E's copy likely pre-qualified buyers up front, filtering out window shoppers.
Variant H degraded performance across every step of the customer journey (32.8% click rate, 14.3% cart rate, 33.7% checkout rate), resulting in 613 fewer purchases than Control A. Variants B, D, and G showed no meaningful difference from A.
Percentage of total landing page visitors who completed a purchase (with 95% confidence intervals). Variant F stands completely alone.
Comparing Top-of-Funnel engagement (Click Rate) against Bottom-of-Funnel qualification (Add-to-Cart to Purchase Rate).
How each redesign altered visitor behavior at each sequential gate. Toggle metrics below to inspect specific drop-off points.
Aggregated totals across all 70 days (June 1 – August 9, 2026). Traffic split is balanced (~227.5k visitors per variant; SRM check p = 0.998).
| Variant | Role / Archetype | Visitors | Clicks | Add to Cart | Purchases | 1. Click Rate (CTR) | 2. Click → Cart | 3. Cart → Buy | Overall Purchase Rate | Lift vs. Control (A) |
|---|
Weekly purchase conversion rate across all 10 weeks. Variant F's advantage is durable from Week 1 to Week 10—not driven by novelty or a single anomaly.
Running total of net additional (or lost) purchases compared to Variant A over the 70-day window. Variant F steadily pulls away every week.
Average conversion rate by day of the week. Traffic peaks on Sundays (~3,510 visitors/variant) and dips on Saturdays (~2,990), with conversion peaking on Saturdays and Mondays.
Concrete action plan based on the 10-week findings.
Variant F is statistically significant well beyond industry standards (Z = +8.55, p < 0.0001) and shows zero novelty decay over 10 weeks. Shipping F to all ~1.82M visitors per 10-week cycle will generate roughly +6,480 incremental purchases per month.
Audit what made Variant E's clickers convert at 40.1% once in the cart (e.g., clearer pricing, specs, or expectation-setting copy) and see if those trust elements can be blended into Variant F without hurting F's 37.9% click rate. Avoid Variant C's aggressive CTA tactics.
Had we judged this test on CTA Click-Through Rate, we would have shipped Variant C—wasting engineering effort for zero sales gain and frustrating users who bounce after clicking.