Performing a Visual Audit: Finding Where Your Interface Repeats and Differs
A visual audit collects every screen, button, color, and font in a product so a team can count what repeats, what differs, and what to fix first.
In short
A visual audit collects every screen, button, color and font in a product so a team can count what repeats, what differs and what to fix first.
12 min readFree
What You Will Learn
This lesson teaches you how to perform a visual audit. A visual audit is a review of how a product looks. You collect every place the interface appears, then record where the styles repeat and where they differ. The end result is an inventory of inconsistencies. A design team can use that inventory to plan design system improvements.
By the end, you should be able to explain why a team runs an audit, walk through the steps, and turn the findings into a short report that sets priorities. You do not need expensive tools. You need screenshots, a clear list of categories, and patience.
Key words
- Visual audit
- A review of how a product looks. You gather every screen and element, then compare them side by side.
- Interface
- Everything a person sees and touches in a product, such as screens, buttons, menus, and forms.
- Inventory
- A complete list of items. In this lesson, it is a list of the visual parts found in a product.
- Scope
- The set of screens or user flows that the audit covers. A flow is a series of screens a person moves through to finish a task.
- Variation
- A version of the same element that looks a little different, such as a button with another color.
- Duplicate
- Two or more copies that look exactly the same.
- Near-duplicate
- Items that look almost the same but differ in one small way, such as a slightly different shade of gray.
- Outlier
- A variation that is used rarely or in only one place, and that does not match the rest.
- Standardize
- To pick one way of doing something and use it everywhere.
Why and When Should You Run a Visual Audit?
A visual audit is usually the first piece of evidence a team gathers when it builds a design system from a product that already exists. A design system is a shared set of parts and rules that a team reuses. If the product was built without one, the audit shows what is already there. It also shows where the team has been making extra, unplanned versions of the same thing.
Many teams start with a feeling. Someone says, "Our interface feels messy." That feeling is hard to act on. An audit turns the feeling into counts and screenshots. Instead of saying the buttons look inconsistent, the team can say how many button styles exist, where each one appears, and how often each one is used.
Inconsistency grows slowly. Small differences build up as a product changes, especially when many people design it. Teams often notice the problem most clearly when the design team is large and different people have worked on different screens.
After years of growth
A product that has been changed by many people over several years often has many near-duplicate styles. An audit finds them in one pass instead of one bug at a time.
Before building new parts
If a team creates new components before it knows what already exists, it may add one more version of something that already has five. An audit comes first so that new parts replace old ones on purpose.
Again, later on
An audit is not a one-time event. A light audit every few releases helps stop small differences from piling up again.
Tip
Audit the product as users see it, not only the design files. Screenshots of the shipped app often show differences that the design files hide.
How Do You Run a Visual Audit, Step by Step?
Most audits follow the same basic path. The method is often called an interface inventory. Brad Frost, an author who wrote about design systems in his book Atomic Design, defined the interface inventory method. The steps below combine his approach with guidance from other design teams.
- 1
Set the scope
Decide which screens or flows you will check. Write the list down. A scope might be the sign-up flow, the checkout flow, and the settings pages. Keep it small enough to finish.
- 2
Capture screenshots
Take a screenshot of every unique screen, dialog box, menu, and empty state in the scope. Use the same tool and the same screen size for every capture so the images can be compared fairly.
- 3
Sort the elements into categories
Group the parts you see into categories such as buttons, colors, type styles, spacing, icons, and form fields. Place each piece of evidence under the category it belongs to.
- 4
Put every variation side by side
For each category, lay out all the versions on one page. Seeing the buttons together is the fastest way to notice that some are copies and some are near-duplicates.
- 5
Count and note the differences
Count how many distinct versions exist in each category. Write down what changes between them, such as color, corner roundness, height, or text size.
The rest of this section explains the first two steps in more detail, and then the last three.
Set the scope and take the screenshots. A good scope is specific. Saying "the whole app" is too big for a first audit. Saying "the checkout flow, from cart to order confirmation" is clear and doable. Once the list is set, capture each screen. Many teams keep one file per product area and name each screenshot after the screen it shows.
Sort and compare the elements. Sorting is the step where the audit becomes useful. Put each element into a category. Do not worry about getting the categories perfect on the first try. Add new categories when you find something that fits none of them. Then, for each category, place all the versions side by side. Notes help here. Write down the details you can read from the screen, such as the font size, the color, the padding (the empty space inside a box around its text), and the corner radius (how round the corners are).
Tip
Capture every screen at the same size, so the screenshots compare fairly. A second pass on a small screen can also reveal quick fixes and workarounds that a large screen hides.
What Categories Should You Check?
The categories below cover most visual audits. You can add more, but these six are a solid starting point for a product that has no design system yet.
Buttons
Check the fill color, text style, corner roundness, height, and width. Check hover and pressed states too, which are the looks a button takes when someone moves the mouse over it or presses it.
Colors
Record every distinct color used for text, backgrounds, borders, and warnings. Group them by the job they do. Many products end up with many grays that look almost identical.
Type styles
List every combination of text size, weight (how thick the letters look), and line height (the space between lines) that appears in headings and body text.
Spacing
Note the gaps and padding that appear between items. Find the ones used most often, and flag the ones that appear only once.
Icons
Collect every icon. Note different sizes, styles, and any icon that means different things on different screens.
Form fields
Record text boxes, drop-down menus, check boxes, and radio buttons. Forms are a common place for small differences to hide.
How Do You Go From Findings to Decisions?
Collecting screenshots is only half the job. An audit is useful only when it leads to action. The second half turns the sorted evidence into choices. The steps below show how to move from a pile of screenshots to a short list of priorities.
- 1
Group the variations
Within each category, put near-duplicates into one group. Ask whether two styles are different on purpose. If the only difference is one slightly different shade of gray, they belong together.
- 2
Spot the most-used versions
Count how often each version appears across the scope. Usage counts can show which version has the most users, and the team should weigh them against its goals.
- 3
Flag the outliers
Mark the versions that appear once or only in an old part of the product. These are not always mistakes, but each one needs a reason to stay. Flag them for review rather than deleting them on the spot.
- 4
Write the summary report
Turn the counts into a short report. The report should name each category, the number of distinct versions, and the most-used choice. Then list the priorities in order.
Check the priority order against the team's goals. A color difference that makes text hard to read is more urgent than a small difference in spacing that nobody can see. Use simple labels, such as critical, major, and minor, to show how serious each issue is. A minor issue might be a slightly different gray. A major issue might be two different styles for the same warning message, which confuses people.
Tip
Keep the audit in one place with three parts: the screenshots, the category inventories, and the decisions. A team can then return to it later without searching through old files.
Write a Simple Report With Clear Numbers
A report should be short enough that a busy person reads it. Counts make it easy to understand. Real audits report counts like these.
Your own report can use the same format. For example: 37 shades of gray, 9 button styles, 4 type combinations used for the same job. Each line should answer three questions. How many versions exist? Which one is used most? What should the system standardize first? A report in this form tells the design system team where to begin.
The count
How many distinct versions exist in each category. This is the headline number.
The most-used choice
The version that appears most, which is the candidate standard.
The priority
Which category causes the most confusion or risk, so the team knows what to standardize first.
Worked Example: Auditing the Buttons in a Checkout Flow
This walkthrough uses a shopping checkout flow as the example. Checkout is the set of screens a shopper moves through to pay, from the cart to the order confirmation. The numbers in this example are made up to show the method. They are not measurements from any real shopping site.
- 1
Set the scope
The scope is the checkout flow: cart, shipping address, payment, order review, and confirmation. Write these five screens on the audit page.
- 2
Capture every button
Screenshot each button on each screen, including buttons in pop-up windows. Save each one with the screen name and a short label, such as "Continue, shipping screen."
- 3
Sort the buttons into one category
Place all the buttons in a row or grid on one page. Put the primary action buttons, such as Buy and Continue, together so they can be compared directly.
- 4
Compare the Buy and Continue buttons
Look at each one closely. Note its fill color, its text, its corner roundness, its height, and its width. In this example, the sheet shows several Continue buttons that look nearly the same but differ slightly in color and corner roundness. Some Buy buttons are also a different size from the others.
- 5
Count the versions
Group the near-duplicates. Suppose the sheet shows four distinct Continue styles and three distinct Buy styles. Record the counts and the screens where each style appears.
- 6
Pick the candidate standard
Find the Continue style that appears on the most screens. Call it the candidate standard, meaning a strong choice to test, not a final decision. Do the same for the Buy style.
- 7
Flag the outliers
A style that appears on only one screen, such as an old confirmation page, is an outlier. Flag it for review. Ask whether the difference is needed or whether it is a leftover.
The result is a clear, small report. It names the button categories, the count for each one, the most-used candidate for each main action, and the outliers to review. The team now knows where to start. It does not yet have a finished button design. That comes later, when the team builds the system from these results.
Tip
Before choosing a candidate standard, check that its text is easy to read on its background. The most common style is not always the best one. The audit shows what is used, not what is right.
Common Mistakes to Avoid
Starting with no scope
If the team tries to audit everything at once, the audit never ends. Pick a clear set of screens first, then expand later if needed.
Counting only the design files
The design files may show one tidy version of a button, while the live product may show more. Capture the product as it runs.
Judging too early
During the capture step, write down what you see but do not yet decide what is right. Decisions come after the full set of evidence is side by side.
Forgetting the reason
Some differences are on purpose, such as a warning button that is red. Ask the team why a variation exists before you flag it.
Leaving the report unread
An audit that no one uses wastes time. Share the short report with the people who will decide what the system standardizes first.
Check Your Understanding
Quick check
A team's app has been built by different designers over three years. Before creating any components, they want to know how inconsistent the buttons and colors really are. What should they do first, and what would they record?
Quick check
An audit of a team's product finds twelve slightly different blue shades. Nine of them appear only in one old settings page. Three appear on most screens. Which step should the team take next?
Sources
- Netguru: How to Run a Design System Audit: A Step-by-Step Guide for UX Teams
- DevSquad: How to Conduct a UI Audit: Benefits, Steps, and Checklist Included
- Ramotion: Design System Audit in Simple Steps
- Design Shack: How to Conduct a Visual Design Audit (And Why You Should)
- Crafttor: How to Run a Design System Audit From Screenshots
- ModeInspect: Design system audit: is your library ready for AI agents?
- 18F Methods: Interface audit
- Atomic Design by Brad Frost, Chapter 4: The Atomic Workflow
- Brad Frost: Interface Inventory
- Smashing Magazine: A Comprehensive Guide To UI Design
- Cubbbix: What is a UI Audit? A Practical Guide to Finding and Fixing Interface Problems
- Marcin Treder (Medium): Design Systems Sprint 1: The Interface Inventory
- Anshul (Medium): Conduct an Interface Audit: Why, When and How to Conduct an Interface Audit
Last updated on 17 November 2026
Related lessons
Learn with me
Every lesson is free. I add new ones as I learn. Have a question, or want help with your own work? Book a short call.