Design Systems · Step 6 · 6 min read
Measuring adoption
Saying everyone uses the system isn't a measure. What to count in design and in code, what Figma's library analytics show and which plans have them, why detaches matter more than inserts, how far behind teams are on versions, and how to read the numbers together with what people tell you.
Adoption isn't whether teams use the system. It's how much of what users see is built with it, and whether the people building it would choose it again. One number can't tell you that. A count of inserts in Figma says designers drag components in; it doesn't say they keep them, or that the code uses them, or that anyone's happy. Measure a few things, from both sides, and always ask why.
What you'll be able to do after reading:
- Choose adoption measures for design, for code and for people
- Read Figma's library analytics, and know which plan they need
- Treat detaches and version lag as signals, not failures
- Show adoption honestly, with example numbers, in a way that leads to decisions
1. Four questions, not one number
| Question | Measure | Where it comes from |
|---|---|---|
| Is it used in design? | Inserts, instances, teams using the library | Figma library analytics |
| Is it kept as it is? | Detaches, compared with inserts | Figma library analytics |
| Is it used in what ships? | Share of screens or components from the system, in code | A scan of the codebase |
| Is it up to date? | Products on the latest major version | Package versions in each product |
| Do people want it? | Satisfaction, questions, contributions | A short survey, the help channel |
That's five rows for four questions, on purpose: "used" has two sides, design and code, and they can disagree. A team can use every Harbour component in Figma and rebuild them all in code.
2. What Figma's library analytics show
Figma's library analytics count, among other things, inserts ("the number of times a component from the library was inserted into a file"), teams using the library, and detaches, logged "any time someone uses the Detach instance setting", with the same for styles and variables. Per component you see the total instances, how many teams use it and in how many files (Figma, View and explore library analytics).
Things to know before you trust them, all from the same page:
- Plan: analytics are on the Education (K-12), Organization, Government and Enterprise plans; the Plans page lists "Design system analytics" for Organization and Enterprise, not Professional (Figma, Plans and features). The Library Analytics API is Enterprise only.
- History: data updates daily and goes back up to a year. Styles and variables have been tracked since 10 October 2024.
- Blind spots: usage in other people's drafts isn't tracked, nested components don't count as insertions, and you only see files in teams open to the organisation.
So a number that looks low might be a team working in drafts. Ask before you worry.
3. Detaches are a question, not a crime
A detach means someone needed the component to do something it doesn't. Sometimes it's a mistake. Often it's research you didn't have to run:
- One component detached by many teams → the component is missing something. Look at what they changed after detaching.
- One team detaching everything → they don't trust the system, or don't know it. Talk to them (Bringing teams onto the system).
- Detaches rising after a release → the release broke something people relied on (Versioning and releases).
Read detaches per 100 inserts, not as a raw count, so a busy month doesn't look like a crisis.
4. A dashboard, with example numbers
Console screens built mostly from Harbour components
Share of screens, 2026, from a quarterly scan of the code (0 to 100%)
- 22%Q1: 22%
- 41%Q2: 41%
- 58%Q3: 58%
- 71%Q4: 71%
Detaches per 100 inserts, Harbour Figma library
Same quarters, from Figma's library analytics (0 to 20)
- 18Q1: 18
- 14Q2: 14
- 9Q3: 9
- 7Q4: 7
The numbers in this figure are made up for this course, to show the shape. Two measures, from two places, each on its own chart: use in code is rising, detaches in Figma are falling. Together they say teams use Harbour and keep it as it is. Alone, either could fool you: high code use with high detaches means the code is right and Figma isn't; low detaches with low code use means designers love it and developers don't build it.
How Harbour measures the code side: once a quarter a script lists the screens of the staff console and counts, for each one, how many of its interface components come from @tidewater/harbour and how many are local. A screen counts as "on Harbour" when most of them do. It's rough, and it's honest about what it counts.
5. People, not only counts
- A short survey, twice a year: how useful is the system, what's missing, would you choose it again. Brad Frost's sample question is a good start: "On a scale from one to five, how useful is the pattern documentation? Any suggestions?" (Brad Frost, Atomic Design, chapter 5).
- The help channel: the most-asked questions are your next docs or your next fix.
- Contributions from outside the system team: the strongest sign that teams think of the system as theirs (Contribution models).
6. Use numbers to decide, not to rank
- Pick a target per measure, per quarter, and a decision it drives: "If the Sailing card's detach rate is still over 15 per 100 next quarter, we redesign it."
- Never rank teams by adoption in public. It turns a signal into a score, and people start gaming it.
- Show the trend, not one snapshot. One quarter means little; four tell a story.
- Label example and estimated numbers as such, always. The value of the system to the business is the next question, and it's in Showing the value of a design system.
7. Common mistakes
- Counting inserts and calling it adoption.
- Measuring Figma only, and never looking at what ships.
- Treating every detach as a failure.
- Comparing teams instead of trends.
- One chart with two scales, so any story can be told.
- Numbers with no decision attached to them.
8. Vocabulary
| Term | Meaning |
|---|---|
| Adoption | How much of what ships is built with the system, and whether people choose it |
| Insert | Placing a library component into a file |
| Detach | Breaking an instance's link to its main component, so it stops getting updates |
| Coverage | The share of screens or components in a product that come from the system |
| Version lag | How far behind the latest release a product is |
| Library analytics | Figma's usage data for a published library |
9. Try it (no tools needed)
- Write the four questions from section 1 for your own system.
- For each, write what you'd count and where you'd get it. Mark the ones you can't measure today.
- For one of them, write a target for next quarter and the decision it would trigger.
Reflection questions: If you could only measure one thing, which would it be, and what would it miss? Who would be upset if you published the numbers, and why?
10. If you already practise
- Look at your ten most-detached components and open three of the detached instances: what did people change?
- Add a code-side measure if you only have Figma's.
- Track version lag per product and put it in every release review.
- Send the twice-yearly survey, and publish what you'll change because of it.
Checklist
Learn more
- Figma Help Center: View and explore library analytics · Figma plans and features (checked November 2026)
- Brad Frost, Atomic Design, chapter 5: Maintaining design systems
Books, courses and videos to go deeper
Book links go to Amazon. As an Amazon Associate I earn from qualifying purchases, at no extra cost to you.
Related
Bringing teams onto the system · Showing the value of a design system · Versioning and releases · Contribution models · Defining success · Design system metrics