A/B Testing Your Forms

A/B testing lets you compare two or more versions of a form to see which one performs better. Instead of guessing whether a different headline, field order, or design will improve your conversion rate, you can run a controlled experiment and let real data decide. A/B testing is available on the Enterprise plan.

What is A/B testing?

A/B testing (also called split testing) is a method of comparing multiple variants of the same form by randomly showing different versions to different visitors. Each visitor sees only one variant, and their behavior is tracked independently. After enough data is collected, you can determine which variant produces the best results: whether that means more completions, faster submission times, or lower drop-off rates.

Creating a test

1. Open the form you want to test in the form editor.
2. Click the A/B Test tab in the top navigation bar.
3. Click Create new test. Your current form automatically becomes Variant A (the control).
4. Click Add variant to create Variant B. This creates a duplicate of your current form that you can modify independently.
5. Make your changes to Variant B. You can add up to four variants total (A, B, C, D) in a single test.
6. Set the traffic split percentages for each variant (see below).
7. Click Start test to begin routing traffic.

Setting traffic split percentages

By default, traffic is split evenly between all variants. If you have two variants, each receives 50% of visitors. With three variants, each receives roughly 33%.

You can adjust these percentages manually using the traffic allocation sliders. For example, if you want to be cautious about a major redesign, you might send 80% of traffic to your proven Variant A and only 20% to the new Variant B. The percentages must add up to 100%.

What to test

You can change anything between variants. Here are the most impactful things to experiment with:

• Field order: Try placing your easiest questions first to build momentum, or lead with the most important question to capture key data early.
• Question wording: Test formal vs. conversational language, or try rephrasing confusing questions that have high drop-off rates.
• Form layout: Compare a single long-scroll form against a multi-page stepped layout to see which feels less overwhelming.
• Visual design: Test different color schemes, font sizes, button styles, or background images to see what resonates with your audience.
• Number of fields: Remove optional fields in one variant to see if a shorter form increases completion rates.
• Call-to-action text: Change the submit button text from "Submit" to "Get my results" or "Send my request" to see if it affects conversions.

Statistical significance

WittyForm uses two statistical methods to determine whether a result is meaningful or just random noise:

• Bayesian inference: Calculates the probability that each variant is the best performer. This updates continuously as new data comes in and provides an intuitive "chance of winning" percentage for each variant.
• Z-test for proportions: A frequentist hypothesis test that calculates a p-value to determine whether the difference between two conversion rates is statistically significant. WittyForm uses a significance threshold of p < 0.05 by default.

Both methods are displayed on the results page so you can interpret them according to your preference. Most users find the Bayesian "probability of being best" easier to understand, while the Z-test provides the traditional statistical rigor that data teams expect.

Reading your results

Navigate to the A/B Test tab to see results while a test is running. The results dashboard shows:

• Visitors per variant: How many unique visitors were routed to each variant.
• Completion rate per variant: The percentage of visitors who submitted the form for each variant.
• Relative improvement: How much better or worse each variant performs compared to the control (Variant A).
• Probability of being best: The Bayesian probability that each variant is the true winner.
• Statistical significance: Whether the Z-test has reached the significance threshold, displayed as a green checkmark or a gray "needs more data" indicator.

WittyForm recommends collecting at least 100 completions per variant before drawing conclusions. The dashboard will display a warning if your sample size is too small for reliable results.

Declaring a winner

Once a test reaches statistical significance, WittyForm highlights the recommended winner with a banner on the results page. You can then click Declare winner next to the winning variant. This does not immediately change your live form: it simply marks the test as concluded and records which variant won.

Applying the winning variant

After declaring a winner, click Apply winning variant to replace your original form with the winning version. This action:

• Sets the winning variant as the primary form that all visitors see.
• Stops routing traffic to other variants.
• Archives the test results for future reference.
• Preserves the original form as a snapshot in your test history so you can revert if needed.

You can also choose Keep running if you want to collect more data, orDiscard test to cancel the experiment and keep your original form unchanged.