This page provides example prompts designed to help you create prompt-based experiments. The goal is to assist you with writing clear, specific, and effective instructions for the AI, so you can achieve the desired visual or functional outcomes when editing or testing a user interface.
- Be specific. Describe the change, target element, and style.
- Provide additional instructions if needed (for example, “Make it uppercase”).
- When unsure, start small and iterate.
If the AI needs more information, it will prompt you to clarify—for example, “How would you like the ‘More details’ button to be changed?”
DO — What you should do
- Be specific with your request
Clearly describe what you want to change or add (for example, “change the button color to dark blue”, “add a delivery icon next to the text”). - Specify where the change should occur
Indicate the exact location (for example, “in the main banner”, “below the headline”, “next to the call-to-action button”). Tip: Use our “Draw a Sketch” feature to circle the element or area you want to update. - Use clear and simple language
Write in a clear, concise, and action-oriented style (for example, “replace the white background with a blue gradient”). - Describe visual elements in detail
Include details like size, color, shape, style, or type of image. You can also import a mockup or design of your variant using the “Import a file” feature. - Choose a pre-defined tool (Optional)
To help you get more targeted results, we’ve designed specific prompt tools like Suggest Ideas, Insert Overlay Element, Change Element Style, etc, available at the bottom of the prompt area. While optional, they can improve the quality and relevance of your output. - Stick to one request per prompt when possible
Avoid stacking multiple unrelated edits in a single prompt. - Use concrete examples if helpful
For example, “add an icon like the Google Maps pin”, “use a blue similar to Facebook’s”. - Include a clear goal or intent behind the variation
For example, “make the button stand out more”, “highlight free shipping”. - Rephrase if the first prompt doesn’t work
Try simplifying or rewording the instruction if you don’t get the result you expected.
If for example with your first prompt, you have the desired output but something is wrong, try adding a comment to your original prompt that fixes the desired output. - Start with a new variation when trying something different
If you want to test a new idea, begin by creating a separate variation. - Use clear prompt language when using a mockup (Add a file feature) and customize for brand identity
When using a file as a template (for popups, banners, etc.) and you want the AI to build something based on it, provide explicit instructions. For example, “Build a newsletter popup by using the attached template file and adapt it to my brand design”. Adding details like “adapt to my brand image” can significantly influence the outcome and ensure that the generated content aligns with your visual identity.
DON’T — What to avoid
- Vague or overly general prompts
Be specific. For example, “make it look nicer” or “change the design a bit”. If you’re looking for inspiration or direction, try using the Suggest Ideas tool available at the bottom of the prompt area.
- Don’t use technical jargon or uncommon abbreviations
Keep it beginner-friendly and visually focused. - Don’t write overly long or confusing prompts
Too many ideas at once make it hard for the AI to respond clearly. - Don’t mix multiple intentions into one prompt
For example, “change the button color, add a promo image, and move the text” — break this up. - Don’t assume the AI understands your marketing goal
Always explain what you’re trying to achieve visually. - Don’t fully rely on the first result
Always doublecheck that the visual change matches what you asked for, and test it in simulation mode. - Don’t use vague terms like “this”, “that area”, or “something cool”
Be specific about what should change and how. - Don’t request changes that require back-end logic
For example, “change the product price”, “adjust delivery options”, or “add a discount for returning customers”.
These types of changes depend on backend systems and business logic, which the AI cannot access or modify. Similarly, requests like “add a loyalty points tracker” or “create a subscription billing flow” involve dynamic data and server-side processing and must be handled by developers.
