browserslist
Incorrect Audience coverage for >=0%
There's some issue with Audience coverage, eg. https://browsersl.ist/#q=%3E%3D+0%25®ion=PL
For >= 0% it shows Audience coverage: 80.0 % which is a surprising result. I expected 100% coverage because I requested all browsers.
For global it's a bit better, yet still incorrect: Audience coverage: 96.7 % https://browsersl.ist/#q=%3E%3D+0%25
Antarctica is only 69.0 %.
I will need help in investigation and fixing.
Can you get Can I Use data manually and do the calculation (there is a probability that data is just broken).
Sure. First I checked PL.js in caniuse-lite and the sum is 86.32653000000002. Then I checked caniuse data and the sum is also 86.32653000000002
I wanted to check history of the file but I noticed it contains "total": 86.32653. I don't know where to find documentation for these fields but instead I found a comment in validator:
// Total amount should be 100 - untracked usage
// If total is too low (especially for US) something is probably wrong
I ran into this issue too!
I haven't gotten to investigate deeply, but I wanted to share some related PRs/Issues I came across, in case it helps someone to investigate.
https://github.com/browserslist/browsersl.ist/issues/423 https://github.com/browserslist/browsersl.ist/issues/512 https://github.com/browserslist/browsersl.ist/pull/513 https://github.com/Fyrd/caniuse/issues/6426 https://github.com/Fyrd/caniuse/issues/6819
I'm also seeing this issue surface when I'm querying for feature coverage
I'm testing with the query cover 98% in US, not dead, not fully supports flexbox
When I run an older version of browserslist that's 9 months old (npx happens to be older), this list looks pretty reasonable, just firefox 11:
npx browserslist "cover 98% in US, not dead, not fully supports flexbox"
Browserslist: browsers data (caniuse-lite) is 9 months old. Please run:
npx update-browserslist-db@latest
Why you should do it regularly: https://github.com/browserslist/update-db#readme
firefox 11
But when I run the same query with today's version of caniuse-lite, I see a much longer list
browserslist "cover 98% in US, not fully supports flexbox, not dead"
android 4.2-4.3
android 4.1
chrome 20
chrome 19
chrome 18
chrome 17
chrome 16
chrome 15
chrome 14
chrome 13
chrome 12
chrome 11
chrome 10
chrome 9
chrome 8
chrome 7
chrome 6
chrome 5
chrome 4
firefox 27
firefox 26
firefox 25
firefox 24
firefox 23
firefox 22
firefox 21
firefox 20
firefox 19
firefox 18
firefox 17
firefox 16
firefox 15
firefox 14
firefox 13
firefox 12
firefox 11
firefox 10
firefox 9
firefox 8
firefox 7
firefox 6
firefox 5
firefox 4
firefox 3.6
firefox 3.5
firefox 3
firefox 2
ios_saf 18.5
ios_saf 6.0-6.1
ios_saf 4.2-4.3
opera 12
opera 11.6
opera 11.5
opera 11.1
opera 11
opera 10.6
opera 10.5
opera 10.0-10.1
opera 9.5-9.6
opera 9
safari 18.5
safari 6
safari 5.1
safari 5
safari 4
safari 3.2
safari 3.1
On the website, I'm seeing that a lot of these have "0.00%" usage:
I've tracked down when the difference in behavior was introduced via the caniuse-lite data:
- April 2, 2025 works as expected, version 1.0.30001709
- April 4, 2025 introduces the audience coverage issue, version 1.0.30001710
I noticed this interesting update around the time of these: MDN was updated around the time of this change. (But this update didn't break the following release, so I don't think this is the issue.) https://github.com/browserslist/caniuse-lite/commit/72210eaffe1efed4d00066582339769391c99430
While this is being worked on
If someone needs a snapshot where audience coverage totals to 100% more reliably, you could use this pnpm override in your package.json:
"pnpm": {
"overrides": {
"caniuse-lite": "1.0.30001709"
}
}
I found the commit where the data change happened that caused this issue: https://github.com/Fyrd/caniuse/commit/d32774345c9c543844f2ef14c9cec5d89d2d6bb8
And some direct links to relevant files' diffs:
Whoops, those links take forever to load!
Here's a direct link to the before and after data for the global / worldwide data set:
And before and after for the US data set:
And before and after for the PL data set: