Mar 10, 2025 · 1,023 views · 4 min read
How I automated metrics in the DS
How AI helps automate the data.
At the end of every month someone on the design system team had to answer the same question: how much is the system actually used? And every month the answer took two days of opening files, counting components and pasting numbers into a spreadsheet.
Two days of manual work for a number everyone looked at for five minutes. Something was off.
First, decide what to measure
Before automating anything, we sat down to agree on what “adoption” meant. We landed on three simple signals: system components versus loose ones, legacy components that should be migrated, and screens that use nothing from the system.
Then, let the machine count
I built a plugin that walks the files and outputs those three signals as JSON. With AI’s help I wrote the script that merges every JSON and generates the report. The AI doesn’t decide what to measure; it just saves me from writing the boring code.
- From two days to fifteen minutes per report.
- Teams see their own number and compare it with last month.
- Conversations moved from “how much is it used?” to “why not here?”.
Automating isn’t about removing people; it’s about giving them their time back to think.
Next up: having the report arrive on its own every Monday. But that’s another post.
