A shorter, conversational pre-lander form increased the reported opt-in rate from 21.12% to 32.05% during an 11-day experiment. That is +10.93 percentage points and a reported 51.77% relative lift. The downstream application display is incomplete, so this page does not claim that the initial lift compounded into a quantified application or sales improvement.
What changed
Visitors arriving before a webinar saw one of two forms. Control used three longer, problem-focused questions in a serious tone. Treatment used shorter questions, quicker answer choices, conversational wording and more emotionally reflective framing. Both forms occupied the same funnel stage before the webinar.
| Dimension | Control | Treatment |
|---|---|---|
| Length | Longer statements and answers | Shorter questions and quick choices |
| Tone | Serious and problem-focused | Conversational and reflective |
| Measured immediate outcome | Pre-lander opt-in | Same pre-lander opt-in |
| Later outcome | Application submission after webinar | Same later event |
Because several dimensions changed together, the experiment estimates the effect of the complete form package. It does not identify tone, length, cognitive effort or emotional framing as the individual cause.
Two metrics, different evidence strength
The immediate opt-in result is reported as 21.12% for control and 32.05% for treatment. The historical display reportedly includes the +51.77% relative comparison, but per-arm assignments and conversions are not available in the page.
For the later application metric, the surviving display shows 2.31% for treatment and a 99.07% win probability while the control rate is missing. Empty interval fields and unavailable denominators prevent a quantified downstream-lift claim. A probability output alone does not show how many applications occurred or whether the event was delivered consistently.
What was not measured
The experiment did not measure attention during the webinar or ask why visitors converted. It therefore does not show that users were more attentive, that tone was “everything,” or that the opt-in change caused a compounding psychological effect. Those are hypotheses for later research.
How to replicate the useful question
Preserve assignment through the funnel and use the same randomized denominator for opt-in, webinar join, application and purchase outcomes. Record eligibility, exposure, duplicate handling and conversion windows. Keep the original question copy so the package can be reproduced; if the goal is mechanism learning, follow with a test that changes one dimension at a time.
An easier first step can increase volume while reducing qualification, so judge the decision with downstream outcomes as well as the initial form. The combined-form case concerns a later system handoff, while this test concerns acquisition before a webinar.
Decision from the available record
The treatment is a strong candidate for this pre-lander based on the reported immediate rate. The downstream story remains unresolved. Recover the application metric configuration, control result and arm counts before claiming a full-funnel win; do not reconstruct missing counts from rounded percentages.
Use the A/B testing metrics guide to predefine the funnel outcomes and the experimentation-system audit to check assignment and event integrity.
