By skyforbes 
ChatGPT is a versatile tool that can be used for various purposes, and it is very handy in product design. Previously, I showed how to use this tool for UI design.
And in this article, I will show another great use case — ChatGPT will audit our UI design and offer practical tips on how to make it better.
What is a design audit and how to do it right?
Design audit is a general term that product teams use when they’re talking about evaluating design according to a certain criterion. Below, I will propose a general workflow for design audit that will cover 6 essential steps.
Step 1: Creating context for ChatGPT
The first thing we need to do is provide instructions for ChatGPT so it will understand its role and the specific actions it should perform. The best way to do it is to start a session with ChatGPT by providing instructions that will clearly articulate the role in the first part of the prompt (i.e., first sentence) and then move to practical tasks ChatGPT should perform. Below is an example of a prompt I will use:
“You are an expert in digital product design responsible for UI audit of [product type] (e.g., medical dashboard, fintech mobile app, SaaS landing page).
I’ll share UI screenshots and design tokens.
And you will create a UI Audit plan with:
- Key heuristics (e.g., Nielsen, Apple HIG, WCAG)
- Audit criteria for visual design, interaction design, accessibility, brand alignment
- Output format: table with [Criterion | Description | Pass/Fail | Notes]”
For my example, I will do audit of a food delivery app I’ve created using ChatGPT 5 (https://uxplanet.org/ui-design-with-chatgpt-5-afc67dc501a1).
So the prompt in my case will be:
“You are an expert in digital product design responsible for UI audit of food delivery services.
I’ll share screenshots.
You will create a UI Audit plan with:
- Key heuristics (e.g., Nielsen, Apple HIG, WCAG)
- Audit criteria for visual design, interaction design, accessibility, brand alignment
- Output format: table with [Criterion | Description | Pass/Fail | Notes]”
So my initial input will look like this:
ChatGPT will likely ask some confirmation questions before doing actual ‘heavy-lifting’ so you will likely see an in-between response like this one below.
In my case, I will go with a focused audit for this particular app. So I will choose A, and ChatGPT will generate the following plan for me.
This plan gives me a general idea of how my design performs (think of it as a birds-eye view), but it’s a bit too general, meaning that it only highlights some areas where we have problems in this UI. So in the next few steps, we will fill this gap.
Step 2: Analyzing layout and visual design
To make our audit more specific, we need to zoom in and break an entire screen into individual sections/components, as well as the styles these components have. This will help us focus on specific design decisions we have in place. To do that, I will submit a follow-up prompt.
“Analyze this UI screenshot and identify:
- Layout structure and hierarchy
- Primary and secondary actions
- Areas of visual tension or inconsistency
- Any missing states (hover, focus, error)
Return findings as structured points under Layout, Typography, Color, Interactions.”
ChatGPT is very good at inspecting images, so it can easily extract not only the text but also the actual font family that the text uses.
Pro tip: When doing a visual audit, you can attach screenshots of a few different screens and ask ChatGPT to evaluate visual consistency—make sure we are using the same design decisions (same components, styles) across our product. Below is a prompt sample for consistency evaluation.
“Use these N screenshots, and identify visual inconsistencies such as misaligned spacing, irregular corner radii, color token mismatches, or inconsistent iconography.
Return:
- List of detected issues
- Suggestions for system-level fixes that will help align consistency”
Step 3: Heuristic analysis
Visual desing inspection is an integral part of product design audit, but it’s only half the battle. The other half is functionality we will offer for our users on this screen. Explanation of in-depth functional analysis requires its own article, but for this article, I want to focus on one important attribute of the functional analysis — heuristic evaluation. Heuristic evaluation is a usability inspection method where we assess a user interface against a set of established principles, or “heuristics,” to find usability problems. It helps us identify potential usability issues before we code the screen.
Here is a prompt sample I will use for heuristic evaluation
“Perform a heuristic evaluation of this interface based on Nielsen’s 10 principles.
For each heuristic, provide:
- Short assessment (1–2 sentences)
- Severity score (0–4)
- Suggested fix.
Format as a markdown table.”
And here is an output that ChatGPT will generate for me:
Similar to visual design analysis, heuristics will give us a more detailed picture of how our design will perform for end users.
Step 4: Accessibility check
Many design audit professionals conduct accessibility checks during step 3 (visual and layout design), but I strongly suggest doing it separately, as accessibility checks are much more than color contrast checks or font size checks. And you will be able to perform a more detailed a11y check when you complete the functionality check of your product (you know the features you offer and how people will use them).
For my example, I will rely on the tried and tested WCAG for this:
“Conduct a quick accessibility audit based on WCAG 2.2:
- Contrast ratio issues
- Tap target sizes
- Text scaling risks
Missing focus indicators or state cues.
Include severity and recommended remediation.”
Once you submit the prompt, you will expect to see something like this:
What I really like about ChatGPT is that it offers a quick summary of this analysis. So I typically read the a11y audit backward — start with a summary and move to the detailed analysis.
Step 5: Emotion and brand design alignment
Emotions play a critical role in product design. The more we align our design with the emotions that users should experience when they’re using our product, the higher the chance we will have a positive experience.
Brand design alignment is something we need to consider when we’re building a product for an established organization. Colors, typography, and imagery are integral attributes of brand language that the org is using to communicate with its users. You submit specific HEX codes or typography choices to ChatGPT as a prompt and then ask to make sure that the design aligns with these choices.
“Analyze this UI’s emotional tone and brand consistency.
Describe:
- Color mood and emotional impact
- Typography personality
- Brand coherence across components
- Does it match the intended brand values: [insert 3 words, e.g., “trustworthy, innovative, calm”]?
Suggest adjustments for alignment.”
Once you submit the prompt, you should expect to see a response like this from ChatGPT.
Step 6: Summary and scorecards
This is the final step of the design audit. During this step, we will ask ChatGPT to create a summary for our audit and make scorecards that we can share with stakeholders.
“Summarize the audit findings as a UI Quality Scorecard.
Use categories:
- Visual consistency
- Accessibility
- Interaction clarity
- Brand coherence
- Design system compliance.
Assign a score (1–10) for each, plus a one-line summary.
Then produce a short executive summary paragraph.”
🚨 New AI-powered product design community 🚨
If you’re passionate about building digital products and want to make the most of latest tools (including AI tools), I’d love for you to join me in the Product Design Community on Skool. You will have access to tutorials and cheatsheets for design and automation tools as well as live Q&A sessions with me.
👉 Join for free here: https://www.skool.com/product-design-7868