esc
  • WorkflowsPages
  • Case studiesPages
  • Machine Learning FoundationsLearn
  • AI EngineeringLearn
  • AI Guardrails & Red TeamingLearn
  • UI/UXLearn
  • Design SystemsLearn
  • Production System DesignLearn
  • Book a callContact
  • Download ResumeContact
Let’s talk

I’ll build the agent for it.

Existing Design Analysis: Find What Your Product Already Has

Before you build a design system, study the product you already ship. Learn what to examine, how to count the mess, and how to set first priorities.

In short

Existing design analysis means studying the product you already ship, counting what is inconsistent and setting first priorities before you build a design system.

11 min readFree

What you will learn

This lesson shows how a team examines its live product before it builds a design system. You will learn what to look at, how to group and count what you find, how to decide what to keep, merge, redesign, or drop, and how to write a short summary that sets the first priorities. Plan about 10 minutes for the read.

Key words

Design system
A shared set of parts and rules that a team uses to design and build products in a consistent way.
Existing design analysis
The step where a team studies what its current product already contains before it builds anything new.
Live product
The version of the app or website that real users see right now, as opposed to the design files behind it.
Interface inventory
A collection of every screen element, style, and pattern found in a product, sorted into groups.
Pattern
A set of parts that work together to solve one common job, such as a sign-up form.
Duplicate
Two or more items that are exactly the same.
Near-duplicate
Items that look almost the same but differ in one small detail, such as height or corner roundness.
Inconsistency
The same kind of thing shown in different ways, when one way would do the job.
Accessibility
Making a product usable by people with disabilities, such as people with low vision.

Why look before you build?

Most teams do not start with an empty page. They already have screens, and many people made those screens over several years. Each person may have picked a slightly different blue or a new button shape because the old one did not fit their task. Nobody plans that kind of mess, but it builds up quietly.

Existing design analysis is the check that comes first. It answers three questions: what already exists, what is inconsistent, and what is worth keeping. The answers show where the system should start and what it must fix.

Tip

This review is a map of the product, not a judgment of the designers who came before. Most drift comes from many small choices made under deadline. Describe what you see, and keep blame out of the notes.

What should you examine?

Start with the live product. A design file shows what someone intended. The running app shows what users actually get, and the two often differ.

  • Screens

    The full set of pages or views, such as search, listing, and checkout. Note the order users take through them.

  • UI elements

    The small parts on those screens: buttons, text fields, cards, menus, tabs, and icons.

  • Styles

    The base choices behind the parts: colors, text sizes, spacing between items, corner shapes, and shadows.

  • Patterns

    Groups of elements that solve one job together, such as a form that collects a shipping address.

  • How they are built

    Whether the design file and the coded product match, and whether the same element exists twice, once in design and once in code.

The last point matters. A design file can look tidy while the product runs something different. Good analysts check both.

How do you build the inventory?

An inventory is everything you found, collected and sorted into groups. The simplest version is a set of screenshots, with one board for each element type: all buttons on one board, all text fields on another, all colors on a third. Crop each example tightly, and write down the screen it came from.

Add one more column: how often each item appears. You will need that count when you make decisions.

Tip

Capture the shipped app, not only the design file. Screenshots of the real product show drift that the file hides.

How do you find patterns and problems?

Once the inventory exists, work through it in a fixed order. Do not judge items while you group them. Judging too early hides the pattern you are looking for.

  1. 1

    Group similar items

    Put every button on one board, then slide the look-alikes together. Each group should be one kind of part, such as primary buttons or text links.

  2. 2

    Count duplicates

    Tally identical items. If the same button appears on many screens, you have one style that is heavily used.

  3. 3

    Count near-duplicates

    Look for items that differ by one small thing, such as height, radius, or shade. For example, fourteen shades of blue is a color problem, not fourteen real design choices.

  4. 4

    Mark inconsistencies

    Note where the same job gets different treatment. Two sign-up forms with different field heights are a simple example.

  5. 5

    Look for gaps

    A gap is a job the product needs but no part handles well, such as error messages that have no agreed style.

  6. 6

    Check accessibility

    Test text contrast, the size of tap targets (the area a finger can press), and whether focus is visible (a clear outline that shows which item is selected) when a user moves with a keyboard.

Check accessibility while you look

Contrast means how far apart the text color and the background color are. The W3C guideline called WCAG (Web Content Accessibility Guidelines) sets the common rule. At level AA, normal text needs a contrast ratio of at least 4.5 to 1 against its background. Large text can go as low as 3 to 1. Large text is generally 18 point or larger, or 14 point and bold, and 18 point is about 24 pixels on screen.

Placeholder text, the faint hint inside an empty field, counts as text, so it needs the same check. Disabled controls are often exempt from the rule, but people still need to read them, so a clear contrast is still the safer choice.

Tip

Check grey secondary text first. Captions, timestamps, and footer lines are where light grey most often fails the contrast rule.

Turn findings into decisions

A finding is a fact about the product. A decision says what to do about that fact. For each group in your inventory, choose one of four outcomes.

  • Keep

    The item is used widely, works well, and fits the product goals. It becomes a base for the system.

  • Merge

    Several versions do the same job. Combine them into one version and retire the rest.

  • Redesign

    The job matters, but the current version fails on usability, accessibility, or clarity. Rebuild it before it enters the system.

  • Drop

    The item is rare, out of date, or does no real job. Remove it, but only after checking that no important flow depends on it.

Be careful with differences. Some variation serves a real purpose, such as a warning button that must look different from a normal one. The goal is to remove confusing differences, not every difference.

How do you rank by use and visibility?

Rank items with two questions. How often is the item used? How visible is it to users? A button on the most common screens, or on the checkout path, matters more than one tucked into a settings page. Add a third question: how inconsistent is it? Items that are used a lot and vary a lot should rise to the top.

Some audit guidance ranks items by how often they are used and how inconsistent they are. Start with the elements that appear most often, because they touch the most screens.

Tip

Ranking is about the first release, not every part of the product. Pick the few items that will reach the most screens, and fix those first.

Worked example: a student reviews a travel app

Here is one way a student could run the method on the Airbnb mobile app. The steps show the method. The findings are what a student would record while working, not a report on the app's real code or internal counts.

Worked example: the steps

  1. 1

    Capture the live screens

    The student takes screenshots from the installed app of the search, listing, and checkout screens. She works from the running app, not from a file.

  2. 2

    Group the buttons

    She puts every button on one board. She sorts them by job: the main action on a screen, a secondary action, and a plain text link.

  3. 3

    Spot the variation

    Within each group, she notices buttons that differ slightly in height, corner roundness, and fill color. Some look like one button shown at different sizes. Others look like copies with small edits.

  4. 4

    Count and tag

    For each version, she notes the screen it appears on, how often it appears, and whether its label has enough contrast against its fill.

  5. 5

    Propose a decision

    She suggests one primary button for the main action and one secondary button for lighter actions. Near-duplicates merge into these two. A version that appears only once is checked, then dropped if no job needs it.

  6. 6

    Write the summary

    She lists the two proposed buttons as the first components. For each, she gives the screens affected, the counts behind the choice, and the contrast check for its label.

Tip

The first components should come from frequent, visible parts. A button used on checkout and search is a stronger start than a rare badge on one settings screen.

Write the summary

The summary is the handoff. It turns a pile of screenshots into a short plan. Keep it to about one page so that busy people will read it. It should answer what exists, what is broken, what the team keeps, and what comes first.

  • Findings

    The groups you examined, the counts of duplicates and near-duplicates, the gaps, and the accessibility problems, each with the screens where it appears.

  • Decisions

    For each group, whether it is kept, merged, redesigned, or dropped, and the one-line reason.

  • First priorities

    The first few components to build, ranked by use and visibility, plus any open questions that still need an answer.

Tip

Ask a teammate who knows the product to check the summary for missing screens before you share it. A missing flow found early is cheap to add.

Common mistakes to avoid

  • Only reading design files

    Files can show a tidy plan while the shipped product drifts away from it. Check the live app too.

  • Counting everything the same

    A button on checkout and a button in a rarely visited menu do not carry equal weight. Rank before you decide.

  • Removing all differences

    Some variation has a purpose, such as a warning style. Separate confusing differences from useful ones.

  • Skipping accessibility

    Low-contrast text in a common part can affect many users at once. Check it during the review, not later.

  • Never finishing

    An audit that never ends delays the first component. Set a scope, finish the summary, and repeat the review later.

Check your understanding

Quick check

A team starting a design system for an app finds 9 different button styles across 30 screens. Two of those styles appear on 80% of the screens. What should the team do with this finding, and why?

Quick check

On a product card, the grey hint text under a price looks faint on a phone. It is about 14 pixels tall and not bold. A teammate says it is fine because the price next to it is dark. What should you do first?

Key takeaways

Existing design analysis looks at what the live product really contains, groups and counts it, and turns the findings into keep, merge, redesign, or drop decisions. It ends with a short summary that names the first components. Start with the parts used most often and seen most, check their accessibility, and keep the summary short enough that people will act on it.

Sources

How is this lesson?

Last updated on 17 November 2026

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.