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.

Tooling Analytics: Measure Whether Design System Tools Help or Slow Teams

Learn what to measure about the tools behind a design system, where the numbers come from, and how to turn them into fixes that work.

In short

Tooling analytics measures whether the tools behind a design system help or slow teams, and turns the numbers into fixes that work.

14 min readFree

What You Will Learn

A design system is more than a set of parts. It also needs tools: a shared Figma library, plugins that help designers, a code package that developers install, checkers that test code, and a documentation site. Tooling analytics is the practice of measuring how people use these tools and how well the tools work. In this lesson you will learn what to measure, where the numbers come from, and how to turn them into a decision that you can test later.

Key words

Tooling
The software and helper programs a design system team makes or relies on, such as a Figma library, a plugin, a code package, or a documentation site.
Figma library
A shared Figma file that publishes parts, such as components and styles, so other design files can use them.
Plugin
A small add-on tool that runs inside a design app and speeds up one kind of task.
Package
A bundle of code that developers download and add to their product. An npm package is one common kind.
Linter
An automatic checker that flags code or style rules that were broken, so people can fix them early.
Insertion
One time someone places a library component into a design file.
Detachment
One time someone breaks the link between a copied component and its library, so it no longer updates with the library.
Adoption
How many people or teams actually use a tool or part.
Friction
Anything that slows people down or pushes them to work around a tool.
Baseline
Your first measurement, kept as a starting point so you can compare later results with it.
Deprecate
Mark a part or tool as old, warn users, and plan its removal.

What does tooling analytics mean?

Tooling analytics means measuring how a design system's own tools are used and how they perform. The goal is simple. You want to see what helps designers and developers, and what slows them down. A tool can look successful on paper and still cause trouble. Many teams count only design-phase numbers, such as how often a component was placed in a file. That count tells you what designers did, not whether the product works well. Good tooling analytics looks at both sides.

  • Design tools

    The Figma library, its components and styles, and any plugins that designers use while working in files.

  • Code tools

    The npm package developers install, the linter that checks their code, and the build pipeline that tests and packages the code each time it changes.

  • Knowledge tools

    The documentation site where people learn how to use the parts and tools.

Tip

Write one sentence that names the tool, the people who use it, and the job it should help them do. If you cannot write that sentence, you are not ready to measure the tool yet.

What should you measure?

Most tooling measures answer three questions. Who uses the tool? What do they use inside it? What gets in their way? Keep these three questions in mind. Each number you collect should answer at least one of them.

How do you measure who uses the tools?

  • Library users

    How many teams insert components from the Figma library, and how their share compares with other teams. Figma's library analytics can show this for each team.

  • Package users

    How many projects import the npm package. Download counts are a start, but they do not show which projects actually run it.

  • Version adoption

    How many teams have moved to the newest version. A team on an old version may not get fixes or new parts.

  • Plugin users

    How many active people use each plugin in a given month. A plugin with only a few users may be unknown or hard to use.

What usage patterns show what is used or ignored?

Adoption tells you who is present. Usage patterns tell you what they do once they are there. Look for the parts of a tool that people use every day and the parts that nobody touches. A pattern that repeats across teams is more useful than a single spike. For example, if one plugin feature is used a lot and the rest of the plugin is ignored, the plugin may be doing one job well and the others poorly.

  • Popular components

    Which library parts are inserted most, and which are rarely placed in any file.

  • Unused features

    Which plugin commands, package options, or linter rules are almost never used.

  • Documentation attention

    Which documentation pages get visits and which search words people type most often.

What friction slows people down?

Friction is the signal that a tool needs work. Some friction shows in numbers, such as slow builds or failed checks. Some shows in behavior, such as people breaking links to library parts. Some shows in requests for help. Each of these can point to a fix.

  • Detachments

    A rising count of detachments can mean a part does not meet a real need, has a bug, or lacks a option that people want.

  • Build times

    How long the build pipeline takes to finish. Slow builds make each change take longer to check.

  • Failed lint checks

    How often the linter stops a change, and which rule causes most of the stops. A rule that fails often may be unclear or hard to follow.

  • Support requests

    Questions and bug reports that trace back to one tool. Count these by tool so the cause is clear.

Be careful with detachments. A detachment does not always mean the component failed. Sometimes the team needed a custom version on purpose. That is why you pair each number with a question and a conversation.

Where the Data Comes From

Tooling analytics uses data from several places. No single place tells the whole story. The next five sub-topics explain the main sources and what each one can and cannot show you.

Design Tool Data

Figma offers built-in library analytics on its Organization, Government, and Enterprise plans, and on some education plans. The dashboard shows how many times components, styles, and variables from a library were inserted, and which teams did the inserting. Each team's share of insertions is shown. The data updates daily, and Figma keeps up to one year of it. Figma added tracking for styles and variables, not only components, starting October 10, 2024.

Read this data with its limits in mind. Figma also offers an API for Enterprise customers, which is a way for other programs to pull this data for custom reports. One team, athenahealth, has used it to combine insertions and detachments into a monthly review.

Tip

Before you trust a dashboard, ask what it leaves out. Write down the limits next to the chart so that the next reader does not assume the chart shows everything.

Package and Version Data

Package data comes from the place where the code bundle is published, plus any usage data the team collects from projects. It shows downloads, how many projects depend on the package, and which version each project uses. Version data is especially useful. It tells you whether teams have moved to the newest release or stayed behind. Netguru lists version adoption as a key measure for design system code.

A download is one event, not proof of use. A download can come from a test, a build server, or a team that never shipped the product. Treat download counts as a rough signal and check them against version and project data.

Build, Lint, and Pipeline Reports

A build pipeline is the automatic chain that checks, tests, and packages code each time someone changes it. Reports from that chain show build durations, how long jobs wait in line, and how often a run passes or fails. Linter reports show which rules fail most often and in which projects. Together they show where developers lose time to the tooling itself.

These reports are objective, but they are not complete. A slow build may be a real problem, or it may be a rare outlier. Compare the numbers over several weeks before you act. Developer experience guides suggest pairing this workflow data with what people say about their work.

Documentation Site Analytics

Documentation analytics show which pages people visit, how long they stay, and what they search for on the site. A page that gets many visits with few follow-up searches may be clear. A search term with no matching page points to a gap. Use the results to guide content and to notice tools that need a better explanation.

Surveys and Conversations

Numbers show what happened. Surveys and short conversations help explain why. Keep a survey short, with a few questions about the tools people use most. Ask about speed, clarity, and what gets in the way. Interviews and quick chats can uncover causes that no dashboard shows.

  1. 1

    Pick a number

    Start with one number that looks odd, such as a rise in detachments of one part.

  2. 2

    Add a question

    Ask the people involved why they made that choice, or what they were trying to do.

  3. 3

    Check the pattern

    Compare what people say with the data. If both point the same way, you have a stronger signal.

Turning Findings Into Decisions

Data is only useful when it leads to a choice. The steps below show a simple cycle. You measure, act, and then measure again to see if the action helped. The steps work for a small team with one library and for a large team with many.

  1. 1

    Set a baseline

    Record where each tool stands before you change anything. Note the date and the method, so the comparison is fair later.

  2. 2

    Pick a few metrics

    Choose three to five measures that match your goal. Too many numbers hide the important ones.

  3. 3

    Spot the problem

    Look for a clear pattern, such as one team stuck on an old version or one part that is detached often.

  4. 4

    Choose the response

    Decide whether to fix a tool, improve it with a new feature or better guidance, or deprecate it and point people to a replacement.

  5. 5

    Check the result

    After a set time, measure the same numbers again and compare them with the baseline. Keep the change only if the numbers and the feedback both improve.

Set a Baseline and Choose Few Metrics

Many teams start by counting everything they can see. That approach creates noise. Pinterest's design systems team, for example, set clear boundaries before measuring adoption. Their measure was a design adoption percentage calculated per file and across the organization.

Athenahealth's design lead suggested a useful habit. Before looking forward, look backward. Using library analytics, you can review a past year, when you already know what happened, and see which signs came before the outcome. Those signs help you pick the metrics to watch next time.

Tip

Write down the baseline and the date next to each metric. A number with no date cannot show that something changed.

Spot Problems and Choose the Fix

Once the baseline is set, look for problems that repeat. The list below shows common patterns and a first response to each. Each response is a choice to test, not a final answer.

  • Teams stuck on old versions

    Compare the versions people use. Squarespace's design systems team used a library comparison filter to watch usage move from old to new versions and to confirm the old one was retired properly.

  • Parts that get detached often

    Athenahealth follows three steps: check the part for obvious issues, reach out to the teams involved, and file tickets for the fixes needed in Figma or in code.

  • Plugins nobody uses

    Ask whether the plugin solves a real task. If it does, fix its workflow. If it does not, consider removing it or pointing people to a better option.

  • Slow or failing checks

    Find the rule or step that causes most of the delay. Improve that step first, then measure build time again.

Quick check

Your design system's npm package was downloaded 500 times last month. The docs site shows most teams still use a version from a year ago, and the linter plugin has only 3 active users. What should you look into first?

Worked Example: Button Detachments

This example follows a team reviewing its shared Figma library. It shows the cycle from data to a change and back to data. The screen named here is Figma's library analytics, which lists each component with its insertions and detachments over the past 30 days.

  1. 1

    Open the component list

    The team opens library analytics for its shared library and switches the filter to components. The table lists each component with its inserts and detaches.

  2. 2

    Spot the outlier

    The Button component shows detachments many times a week. Its detachments are high compared with its insertions, so the team marks it for a closer look.

  3. 3

    Ask why

    The team talks with designers who detached the button. They find that many want a button style that the library does not offer. Some are using a manual workaround.

  4. 4

    Check the helper

    The team learns that a plugin is meant to apply button variants quickly. Its usage numbers are low. The plugin exists to solve this exact problem, but people do not use it.

  5. 5

    Fix the workflow

    The team simplifies the plugin's steps, so a designer can pick a variant in one action. They announce the change in the team channel and in the documentation.

  6. 6

    Watch the result

    The next month, the team compares detachments with the baseline. Detachments drop, and plugin use rises. The team keeps the change and keeps watching for any new spike.

Tip

A detachment is a clue, not a verdict. The button example worked because the team asked people why before it changed anything.

Common Traps

Tooling analytics is easy to get wrong. The traps below appear often in real teams. Knowing them early saves time and protects trust in the data.

  • Counting insertions as adoption

    Insertions show that designers placed parts in files. They do not show that the product uses the system well. Pair them with a code-side or product-side measure.

  • Treating a high number as proof

    A dashboard can look healthy while people keep a private channel of workarounds. One team reported that a high library connection rate hid many complaints from developers.

  • Reading every detachment as failure

    A high detachment rate can mean a part was built for custom use and is working as planned. Check the context before you call it a problem.

  • Measuring without acting

    A report that no one reviews changes nothing. Set a review date and name an owner for each metric.

Quick check

Your team's Figma dashboard shows that almost every designer has the library connected and insertions are rising. In the same month, developers say the coded buttons break often and they keep building their own. What is the best reading of this data?

Quick Summary

Tooling analytics measures how design system tools are adopted, used, and experienced. Measure who uses each tool, what they use, and what slows them down. Collect data from design tools, packages, build and lint reports, documentation pages, and short surveys. Set a baseline, watch a few metrics, and use the data to choose a fix, an improvement, or a deprecation. Then measure again to see if the change worked. Numbers start the conversation. People's reasons finish 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.