I’ve been asking myself on one of my latest projects, “How do I validate this darn prototype, but without spending so much time upfront?”.
I don’t know exactly where to start, however, some things that I do know are:
- I need to figure out why users would want to use this
- I need to understand where my current prototype falls down
- I need to know where to go next
After asking myself these questions, I decided to speak with my good friend, ChatGPT.
Here’s one of the prompts that I built out to help him help me figure this out:
You are a UX Feeling specialist. Evaluate the emotional atmosphere
this design creates by assessing color psychology, typography impact,
imagery resonance, and overall mood. Provide:
• One question: What feelings does this evoke?
• A 0–100 score with rating
• A brief rationale linking elements to the score
• Two specific improvement bullets
I then attached an image and hit send!
It’ll typically provide some sort of useful data, for example, I tried this prompt with our homepage
And the here’s the response!
Question
• What feelings does this evoke when you first land on the page?Score: 85 /100 (Very Positive)
Rationale
• Color psychology: The vibrant purple-to-pink gradient feels creative, optimistic, and futuristic, which primes users for an engaging, forward-looking experience.
• Typography impact: The bold, rounded headline type feels friendly and confident, reinforcing approachability.
• Imagery resonance: The playful astronaut mascot adds whimsy and humanizes the tech, making the interface feel more fun than clinical.
• Overall mood: A sense of energetic exploration and safety—users feel invited to “discover” rather than be tested.Improvement Suggestions
• Refine background details: Tone down or remove some of the floating stars/cloud elements in the hero to reduce visual noise and help the core message and CTAs stand out more.
• Increase CTA contrast: Darken the border or add a subtle drop shadow to the “Book a Demo” button so it pops more against the vibrant background and draws clearer attention.
As you can see, there are definitely some good pointers in here!
The real question is, how can we make the prompt even better? What sort of context can we add to make the feedback more powerful?
Would love to see if anyone else has been messing with this!
