This experiment tested whether changing the primary call-to-action button text from "Get Started" to "Start Free Trial" would increase conversion rates on our homepage. After 21 days and nearly 50,000 visitors, we observed a 27.4% lift in conversions with 96.4% statistical confidence—well above our predetermined 95% threshold for declaring a winner.
The "Start Free Trial" variant consistently outperformed the control across all user segments, device types, and traffic sources. Based on these results, we recommend implementing Variant B as the new default and projecting an additional ~3,200 monthly sign-ups based on current traffic levels.
Why We Ran This Test
Our homepage serves as the primary entry point for new users, with the CTA button being the most critical conversion element. Despite strong organic traffic growth (up 34% YoY), our sign-up conversion rate had plateaued at 12.4% for the past two quarters. Qualitative research through user interviews revealed a common theme: potential users were uncertain about what "Get Started" actually meant—would they be charged immediately? Would they need to provide payment information?
The phrase "Start Free Trial" was hypothesized to address this friction by explicitly communicating two key value propositions: (1) there's no immediate cost, and (2) users can try the product before committing. This aligns with research from the Baymard Institute showing that explicit "free trial" messaging can increase conversion rates by 10-30% in SaaS contexts.
- User interviews (n=12) revealed confusion about pricing expectations
- Heatmap analysis showed 23% of users hovering over CTA without clicking
- Competitor analysis: 7 of 10 top competitors use "free trial" language
Test Design & Implementation
We implemented a classic A/B test using a 50/50 traffic split with consistent assignment based on visitor cookies. Users were bucketed on their first visit and remained in the same variant throughout their session and subsequent visits. The test was deployed using our internal experimentation platform with server-side rendering to eliminate flickering.
Primary Metric
Sign-up conversion rate: Percentage of unique homepage visitors who complete account registration within 7 days of first visit. We chose a 7-day attribution window to capture users who research before converting.
Guardrail Metrics
Trial-to-paid conversion: Ensuring we don't attract lower-quality sign-ups. Time to first value: Monitoring if expectation mismatch causes early drop-off.
With a baseline conversion rate of 12.4%, we calculated that detecting a 15% relative lift (minimum detectable effect) with 80% power and 95% confidence would require approximately 21,000 visitors per variant. We planned for 3 weeks of runtime to achieve 24,000+ per variant.
Get Started (Control)
Start Free Trial
Primary Results
Variant B ("Start Free Trial") achieved a conversion rate of 15.8% compared to the control's 12.4%, representing a 27.4% relative improvement. This difference is statistically significant with a p-value of 0.018 (well below our 0.05 threshold) and a 96.4% confidence level.
Conversion Funnel
Funnel Analysis
The most significant improvement occurred at the CTA click stage, where Variant B saw a 25% increase in click-through rate (44% vs 35% of visitors). This validates our hypothesis that the original "Get Started" copy was causing hesitation. Interestingly, the form completion rate also improved by 8%, suggesting that users who clicked with clearer expectations were more committed to completing registration.
Segment Performance
Quality Validation
A critical concern with "free trial" messaging is whether it might attract lower-quality users who never intend to convert to paid plans. To monitor this, we tracked trial-to-paid conversion rates and early engagement metrics for both variants.
All guardrail metrics show neutral-to-positive movement. The slight improvement in trial-to-paid rate (+1.3%) and 7-day retention (+2.8%) suggests that setting clearer expectations actually attracts more committed users, not less. This aligns with the "expectation setting" principle in conversion optimization literature.
Why Did This Work?
The success of this test can be attributed to several psychological principles:
Loss Aversion Reduction
"Free" explicitly signals no financial risk, addressing the primary concern users have when considering a new product. The word "trial" further emphasizes the low-commitment nature of the action.
Cognitive Clarity
"Get Started" is ambiguous—started with what? "Start Free Trial" provides a clear mental model of what happens next, reducing the cognitive load required to make a decision.
Social Proof Alignment
Users are accustomed to "free trial" patterns from other SaaS products. Using familiar language leverages existing mental models and reduces perceived novelty risk.
Expectation Setting
By being explicit about the trial nature, we attract users who are genuinely interested in evaluating the product, leading to higher-quality sign-ups and better downstream metrics.
Limitations & Caveats
While the results are statistically significant and practically meaningful, several limitations should be noted:
- Seasonality: The test ran in January, which may have different user behavior than other months. We plan to monitor post-implementation to ensure results hold.
- Long-term effects: Trial-to-paid data is based on early cohorts. Full validation requires 2-3 months of post-implementation tracking.
- Single-page scope: This test only modified the homepage CTA. Applying similar changes to other pages may yield different results.
Ready to Call Winner
Variant B ("Start Free Trial") has achieved statistical significance with 96.4% confidence. The 27.4% lift in conversion rate exceeds the target 15% improvement threshold.
Recommended Actions
Implement on Homepage
Deploy Variant B as the new default for all homepage visitors. Target completion: Feb 15.
Extend to Other Pages
Design follow-up tests for pricing page, feature pages, and blog CTAs using similar "free trial" messaging.
Monitor Long-term
Track trial-to-paid conversion for 90 days post-implementation to validate quality metrics at scale.
Test Button Design
Now that copy is optimized, explore visual variations (color, size, placement) for additional gains.