Set Up a Synthetic A/B Test in Wingify

Overview

Synthetic A/B testing in Wingify enables teams to validate design hypotheses instantly using Wandz AI-generated personas before committing live web traffic to a standard A/B test.

These personas are artificial, AI-generated users. They're designed to evaluate your control and variation pages the way real visitors might. Each persona reacts based on its defined traits and goals, responding to layout, messaging, and design choices.

By uploading screenshots of your control and variation pages, you can simulate user feedback. This feedback comes from customized synthetic personas and evaluation questions. It helps you assess early directional performance without setup wait times.

Prerequisites

• An active Wandz Advanced account.
• Control and variation page screenshots in SVG, PNG, or JPG format.

Create a Synthetic A/B Test

Creating a test involves the following sequential steps:

• Step 1: Upload Screenshots
• Step 2: Select or Create Target Personas
• Step 3: Define Evaluation Questions
• Step 4: Configure Participant Volume
• Step 5: Review and Launch

You can stop at any point, and Wingify automatically saves your progress as a draft in the Synthetic A/B dashboard. The dashboard indicates the next step you need to complete.

Step 1: Upload Screenshots

1. In the left navigation bar, navigate to Wandz > Synthetic A/B.
2. On the Synthetic A/B dashboard, click Create test.
3. In the setup wizard under the Screenshots tab:
   1. Click Upload Control Screenshot (or drag and drop) to add screenshots of your control (baseline) design.
   2. Click Upload Variation Screenshot (or drag and drop) to add screenshots of the proposed design.
4. Click Continue: Select Persona.

Step 2: Select or Create Target Personas

Personas represent synthetic end-user profiles with defined backgrounds, goals, and traits. Wingify uses these personas to generate synthetic or AI-generated participants who evaluate your uploaded screenshots and answer the evaluation questions you provide.

1. Select one or more target personas from the Suggested Personas list (for example, UX Insight Specialist, Growth Marketer, or Price-Sensitive Shopper). There's no limit to how many personas you can select. Choose as many as are relevant to the audience segments you want feedback from.
2. Optionally, to build your own, click Custom Persona at the bottom of the window and add the following details:
   • Persona Name: A short label (for example, Tech Professional).
   • Description: The persona's background and goals (for example, A tech professional who stays current with the latest trends and tools in technology).
   • Key Traits: Comma-separated attributes (for example, Curious, Analytical, Tech-savvy).

3. Click Create to save the persona.

   Note: Creating a new persona costs more credits than reusing an existing one.
4. Click Continue: Add questions.

Step 3: Define Evaluation Questions

Choose questions that the synthetic participants should answer when comparing your screenshots. Wandz uses the participants' answers here to decide the winning variation. So, pick the questions that matter most for what you're testing.

1. Select from the recommended question prompts:
   • What is your first impression of this page?
   • What would you click on first?
   • What confused you about this design?
   • How would you complete your main goal here?
2. To add your own question, click Custom Question, type your text into the field, and confirm.
3. Click Continue: Select no of participants.

Step 4: Configure Participant Volume

Set how many synthetic AI participants will evaluate your test using the following steps:

1. Adjust the count using the numeric input or the + / − buttons.

2. The panel shows recommended participant ranges to help you pick. Use a smaller count (1–25) for quick validation, and a larger one for deeper analysis on higher-impact pages.

   Note: Higher participant counts increase credit usage. For example, a test with 20 existing personas costs approximately 50 credits. If the personas need to be created first, the test costs approximately 80 credits.
3. Click Review Test Setup.

Step 5: Review and Launch

1. On the Summary view, review your uploaded screenshots, selected personas, evaluation questions, and participant count.
2. To edit any section, click the pencil icon next to its header.
3. Click Launch Synthetic A/B Test.

The test transitions to a running state. Results typically populate within a few minutes, depending on the participant volume.

View and Analyze Reports

Once a test completes, the results are stored permanently in your account.

To view the reports:

1. Go to Wandz > Synthetic A/B.
2. Find your test in the list and check the Decision column for the outcome (for example, Variation wins, 97% confidence).
3. Click into the test to open the Analysis summary. This shows the winning variation, how many personas chose it (for example, Chosen by 8 of 8 personas), a summary explanation of why it won, and a Key Findings section.
4. Click View Full Analysis to view:

• Side-by-side comparison: Screenshots of your control and variation pages.
• Preference split: Exact ratios and percentages (for example, Variation Preference: 100.0% - 8 out of 8 personas).
• Individual feedback: Detailed participant insights, including their assigned persona and evaluated traits.

There are no further actions available from a completed test. There is no direct path to launch it as a live campaign. To validate a winning hypothesis with real traffic, create a standard campaign in Wingify Web Experimentation.

Manage Test Drafts and Filters

Use the Synthetic A/B dashboard to track progress, resume drafts, and filter your test results.

• Drafts: If you exit before launching, your progress is saved automatically. The Progress column on the Synthetic A/B dashboard indicates completed steps and the next required action (for example, 3 of 5 - Next: Set participants).
• Filters: Use the Status dropdown to filter between Draft and Completed tests, or click Add filter to sort by Participants or Created at.

FAQs

• Does running a Synthetic A/B test consume Wandz AI credits?

  Yes. Synthetic A/B test executions draw from your allocated Wandz AI account credits. Credit usage depends mainly on the number of personas you run and whether you create new ones or reuse existing ones. For more information about credit usage, see Manage Wandz AI Credits.

• Does Synthetic A/B support multiple variations or multivariate testing? 

  No. Synthetic A/B testing is limited to one Control and one Variation. It does not support tests with more than one variation, multivariate (MVT) testing, or Split URL testing.

• How long are Synthetic A/B test reports stored in Wingify?

  Test data and generated analytical reports remain accessible in your account indefinitely.

• How accurate are Synthetic A/B results?

  There is currently no evaluation mechanism or published accuracy benchmark for measuring how closely synthetic A/B responses reflect real-user behavior. Treat Synthetic A/B results as directional input for validating hypotheses, not as a substitute for live user testing.

• Do personas account for regional, device, OS, or platform differences?

  No. Personas include attributes such as nationality and residence, but these attributes are not currently used to model regional, device, OS, or platform-specific behavior. Any regional or platform-specific patterns that appear in the results are incidental.

Need more help?

For further assistance or more information, contact Wingify Support.