# Fedora usage stats

**URL:** <https://discussion.fedoraproject.org/t/fedora-usage-stats/144635>\
**Category:** Project Discussion\
**Tags:** commops-team, marketing-team, kinoite-team, design-team, idea, engineering, mindshare, websites-and-apps-team\
**Created:** [February 14, 2025, 1:54am UTC](https://discussion.fedoraproject.org/t/fedora-usage-stats/144635 "2025-02-14T01:54:46Z")\
**Posts on this page:** 20\
**Page:** 1

<div class="post-metadata">

**Author:** ![nikodunk](https://sea1.discourse-cdn.com/fedoraproject/user_avatar/discussion.fedoraproject.org/nikodunk/32/4712_2.png) [@nikodunk](https://discussion.fedoraproject.org/u/nikodunk)\
**Post date:** [February 14, 2025, 1:54am UTC](https://discussion.fedoraproject.org/t/fedora-usage-stats/144635/1 "2025-02-14T01:54:46Z")

</div>

I’ve always enjoyed @mattdm ‘s Flock presentations on the growth/status of Fedora (and now with matt moving on this is even more timely!). I was just enjoying @castrojo’s [stats on universal blue](https://universal-blue.discourse.group/t/just-how-many-of-us-are-there/6852) (downstream atomic images of Fedora), and was wondering if we could present our Fedora countme stats (for those who haven’t disabled it) in an easily digestible and publicly viewable graph like this.

Without this, the closest publicly available usage data that most people know about are Steams OS leaderboards, and distrowatch. Both not super optimal.

Or does this already exist and I haven’t found it? Thank you everyone!

---

<div class="post-metadata">

**Author:** ![mattdm](https://sea1.discourse-cdn.com/fedoraproject/user_avatar/discussion.fedoraproject.org/mattdm/32/2450_2.png) [@mattdm](https://discussion.fedoraproject.org/u/mattdm)\
**Post date:** [February 14, 2025, 3:30pm UTC](https://discussion.fedoraproject.org/t/fedora-usage-stats/144635/2 "2025-02-14T15:30:25Z")

</div>

Yeah, @gwmngilfen and I are working on this. That and _even better_. 🙂

---

<div class="post-metadata">

**Author:** ![nikodunk](https://sea1.discourse-cdn.com/fedoraproject/user_avatar/discussion.fedoraproject.org/nikodunk/32/4712_2.png) [@nikodunk](https://discussion.fedoraproject.org/u/nikodunk)\
**Post date:** [February 14, 2025, 3:57pm UTC](https://discussion.fedoraproject.org/t/fedora-usage-stats/144635/3 "2025-02-14T15:57:04Z")

</div>

Amazing! Thank you!!

---

<div class="post-metadata">

**Author:** ![popey](https://sea1.discourse-cdn.com/fedoraproject/user_avatar/discussion.fedoraproject.org/popey/32/42837_2.png) [@popey](https://discussion.fedoraproject.org/u/popey)\
**Post date:** [February 14, 2025, 4:01pm UTC](https://discussion.fedoraproject.org/t/fedora-usage-stats/144635/4 "2025-02-14T16:01:30Z")

</div>

> [@nikodunk](#):
>
> distrowatch

Not only is distrowatch “not super optimal”, it’s not remotely close to measuring actual users. It’s just clicks on a page. Don’t make me [tap the sign](https://popey.com/blog/2021/01/distrowatch-is-not-a-measure-of-popularity/) 😃

---

<div class="post-metadata">

**Author:** ![gwmngilfen](https://sea1.discourse-cdn.com/fedoraproject/user_avatar/discussion.fedoraproject.org/gwmngilfen/32/3029_2.png) [@gwmngilfen](https://discussion.fedoraproject.org/u/gwmngilfen)\
**Post date:** [February 14, 2025, 4:13pm UTC](https://discussion.fedoraproject.org/t/fedora-usage-stats/144635/5 "2025-02-14T16:13:12Z")

</div>

Wait, why do I feel I’ve just been volun-told about this … 🤣

But yeah, I’d love to do some nice data viz. Fedora has tons of interesting and unique data, and we absolutely should showcase it.

---

<div class="post-metadata">

**Author:** ![nikodunk](https://sea1.discourse-cdn.com/fedoraproject/user_avatar/discussion.fedoraproject.org/nikodunk/32/4712_2.png) [@nikodunk](https://discussion.fedoraproject.org/u/nikodunk)\
**Post date:** [February 14, 2025, 4:14pm UTC](https://discussion.fedoraproject.org/t/fedora-usage-stats/144635/6 "2025-02-14T16:14:44Z")

</div>

@popey Yeah I was trying to be polite 🙂 Steam’s survey also only reflect gamers, on steam, who were surveyed.

And yet when no one has any info to go off, these are the sources that are quoted.

Now contrast that with the [universal blue post](https://universal-blue.discourse.group/t/just-how-many-of-us-are-there/6852) above: Bazzite has 27K weekly active users, and Fedora atomic desktops have roughly 24k combined (Silverblue, kinoite, aurora, bluefin). 50-ish k total for image mode. Now those are numbers you can take to an app or game developer and hold some water!

---

<div class="post-metadata">

**Author:** ![nikodunk](https://sea1.discourse-cdn.com/fedoraproject/user_avatar/discussion.fedoraproject.org/nikodunk/32/4712_2.png) [@nikodunk](https://discussion.fedoraproject.org/u/nikodunk)\
**Post date:** [February 14, 2025, 6:24pm UTC](https://discussion.fedoraproject.org/t/fedora-usage-stats/144635/7 "2025-02-14T18:24:23Z")

</div>

If helpful @mattdm @gwmngilfen , it appears u-blue runs this automated script in a CI once a day that auto-visualizes the csv from [https://data-analysis.fedoraproject.org](https://data-analysis.fedoraproject.org)

- the repo: [GitHub - ublue-os/countme: countme](https://github.com/ublue-os/countme)
- ci runner: [Workflow runs · ublue-os/countme · GitHub](https://github.com/ublue-os/countme/actions)
- the job: [countme/.github/workflows/countme.yml at main · ublue-os/countme · GitHub](https://github.com/ublue-os/countme/blob/main/.github/workflows/countme.yml)
- python script the job runs: [countme/countme.py at main · ublue-os/countme · GitHub](https://github.com/ublue-os/countme/blob/main/countme.py)
- output: [countme/growth\_global.svg at main · ublue-os/countme · GitHub](https://github.com/ublue-os/countme/blob/main/growth_global.svg)

EDIT: I do see there’s something similar here already, though I’m slightly confused as to what I’m looking at? [Index of /csv-reports/images](https://data-analysis.fedoraproject.org/csv-reports/images/). I gather from [this](https://data-analysis.fedoraproject.org/csv-reports/images/fedora-stacked.png) fedora hit about 500K installs before something happened in F40 and we stopped counting?

---

<div class="post-metadata">

**Author:** ![popey](https://sea1.discourse-cdn.com/fedoraproject/user_avatar/discussion.fedoraproject.org/popey/32/42837_2.png) [@popey](https://discussion.fedoraproject.org/u/popey)\
**Post date:** [February 14, 2025, 10:36pm UTC](https://discussion.fedoraproject.org/t/fedora-usage-stats/144635/8 "2025-02-14T22:36:08Z")

</div>

Here’s some fun and wild maths for a Friday evening. 🤓 Feel free to ignore everything in this post, even though it’s based on actual data with added wild speculation on top :D.

> **It's madness below here**
>
> I just dug up some historical data (I am a data packrat) . I can’t reveal the source, sorry. In 2019, a very, _very_ popular generic app used by nerds and normal people alike had this proportion of installs:
> 
> | Distribution | Percentage |
> | --- | --- |
> | Ubuntu 18.04.2 LTS | 27.2% |
> | Ubuntu 16.04.6 LTS | 10.1% |
> | Linux Mint 19.1 | 5.6% |
> | Ubuntu 18.04.1 LTS | 5.5% |
> | Ubuntu 16.04.5 LTS | ~4.7% |
> | Linux Mint 18.3 | ~4.2% |
> | Ubuntu 19.04 | ~3.8% |
> | Linux Mint 19 | ~3.3% |
> | Debian GNU/Linux 9 (stretch) | ~2.8% |
> | Ubuntu 14.04.5 LTS | ~2.3% |
> | Ubuntu 18.10 | ~2.0% |
> | Ubuntu 16.04.3 LTS | ~1.8% |
> | Ubuntu 16.04.4 LTS | ~1.5% |
> | Arch Linux | ~1.5% |
> | Manjaro Linux | ~1.2% |
> | Ubuntu 14.04.6 LTS | ~1.0% |
> | Remaining distributions | ~22.5% |
> 
> Note: All percentages after the first four are approximations based on visual interpretation of a graph, not raw data. The remaining distributions (adding up to roughly 22.5%) each represent less than 1% of the total and include various versions of Fedora, elementary OS, Deepin, openSUSE, other Ubuntu versions, KDE neon, Gentoo, and Kali Linux.
> 
> Back then I had access to some other data sources, and was able to corroborate some of this. The hand-wavy numbers are in the right ball-park, based on the data I saw.
> 
> Let’s do some aggregation:
> 
> First, we can aggregate the distributions by their base family:
> 
> | Distribution Family | Total Percentage | Breakdown |
> | --- | --- | --- |
> | Ubuntu | ~56.9% | 18.04.2 LTS (27.2%), 16.04.6 LTS (10.1%), 18.04.1 LTS (5.5%), 16.04.5 LTS (4.7%), 19.04 (3.8%), 18.10 (2.0%), 16.04.3 LTS (1.8%), 16.04.4 LTS (1.5%), 14.04.6 LTS (1.0%) |
> | Linux Mint | ~13.1% | 19.1 (5.6%), 18.3 (4.2%), 19 (3.3%) |
> | Debian | ~4.3% | Debian 9 Stretch (2.8%), Debian 8 Jessie + Debian 10 Buster (~1.5%) |
> | Arch-based | ~2.7% | Arch Linux (1.5%), Manjaro (1.2%) |
> | Fedora | ~2.5% | Various Workstation Editions (27-30) |
> | elementary OS | ~1.8% | elementary OS 5.0 Juno + 0.4.1 Loki |
> | openSUSE | ~1.5% | Leap + Tumbleweed versions |
> | Other | ~17.2% | Including KDE neon, Gentoo, Kali, Deepin, CentOS, and others |
> 
> Some interesting observations:
> 
> 1. Ubuntu and its direct derivatives (including Mint) account for about 70% of all installations
> 2. Debian and Debian-based distributions (including Ubuntu and its derivatives) represent over 74% of all installations
> 3. Rolling release distributions (like Arch, Manjaro) make up a relatively small percentage
> 4. The “Other” category shows significant fragmentation in the Linux ecosystem, with many distributions having small but dedicated user bases
> 
> Now for the madness. Let’s take the graph @nikodunk just posted, and use the (estimated) numbers for contemporary Fedora releases, and wildly extrapolate to the other percentages.
> 
> 1. Looking at Fedora in the graph around May 2019:
> 
> - Fedora 29/30 appears to be around 300K unique IPs
> - From our previous analysis, Fedora represented ~2.5% of the total Linux distribution share
> 
> 1. If we use this as a reference point, we can attempt to approximate total numbers:
> 
> - If 300K = 2.5%, then 100% would be approximately 12 million unique IPs
> 
> Let’s wildly plug numbers in and calculate the approximate numbers for the major distributions based on our previous percentages:
> 
> | Distribution | Percentage | Estimated Unique IPs |
> | --- | --- | --- |
> | Ubuntu (all versions) | 56.9% | ~6.8M |
> | Linux Mint | 13.1% | ~1.6M |
> | Debian | 4.3% | ~520K |
> | Arch-based | 2.7% | ~320K |
> | Fedora | 2.5% | ~300K (our reference point) |
> | elementary OS | 1.8% | ~220K |
> | openSUSE | 1.5% | ~180K |
> | Other | 17.2% | ~2.1M |
> 
> However, I should note several important caveats:
> 
> 1. The Fedora numbers are from Yum unique IPs, which might not directly correlate with actual installations
> 2. Different distributions might have different reporting mechanisms
> 3. Some users might disable reporting
> 4. IP addresses can change, leading to overcounting
> 5. NAT and shared IPs can lead to undercounting
> 
> Now let’s dial the madness up to 11, and extrapolate forward _all_ distros, if they grew at the same rate as the Fedora graph (of real yum data) shows.
> 
> Looking at the graph around early 2024 (the end of the data), it appears Fedora’s numbers have increased to approximately 450-500K unique IPs, representing roughly a 50-67% increase from the 300K in 2019.
> 
> Using the higher estimate (500K = 67% increase) and maintaining the same proportions, here’s what we’d get:
> 
> | Distribution | 2019 Est. | 2024 Est. (+67%) |
> | --- | --- | --- |
> | Ubuntu (all versions) | ~6.8M | ~11.4M |
> | Linux Mint | ~1.6M | ~2.7M |
> | Debian | ~520K | ~870K |
> | Arch-based | ~320K | ~535K |
> | Fedora | ~300K | ~500K (reference point) |
> | elementary OS | ~220K | ~367K |
> | openSUSE | ~180K | ~300K |
> | Other | ~2.1M | ~3.5M |
> | **Total** | ~12M | ~20M |
> 
> However, this comes with several important caveats:
> 
> 1. Different distributions likely have different growth rates
> 2. The rise of containers, cloud computing, and WSL might affect these numbers differently across distributions
> 3. The methodology of counting unique IPs might have changed over time
> 4. Some distributions might have gained or lost relative market share
> 5. None of this takes into account new distros (such as Pop\_OS! and other popular Arch builds)
> 
> It’s fun, and very wild speculation. But still better than distrowatch numbers. 😃

---

<div class="post-metadata">

**Author:** ![nikodunk](https://sea1.discourse-cdn.com/fedoraproject/user_avatar/discussion.fedoraproject.org/nikodunk/32/4712_2.png) [@nikodunk](https://discussion.fedoraproject.org/u/nikodunk)\
**Post date:** [February 15, 2025, 12:10am UTC](https://discussion.fedoraproject.org/t/fedora-usage-stats/144635/9 "2025-02-15T00:10:11Z")

</div>

How FUN! Great speculation and reasoning - thank you for sharing and all the work collating this! 😃

---

<div class="post-metadata">

**Author:** ![nikodunk](https://sea1.discourse-cdn.com/fedoraproject/user_avatar/discussion.fedoraproject.org/nikodunk/32/4712_2.png) [@nikodunk](https://discussion.fedoraproject.org/u/nikodunk)\
**Post date:** [May 12, 2025, 8:09pm UTC](https://discussion.fedoraproject.org/t/fedora-usage-stats/144635/10 "2025-05-12T20:09:19Z")

</div>

I don’t know if this is correct, but using the [Universal Blue’s weekly active user](https://github.com/ublue-os/countme) counting methodology running on the `countme` stats, I believe of those 500K IPs about 200K Workstations check in weekly @mattdm @popey.

This includes an adjustment _down_ from a bug @siosm discovered this morning though (which lead me down this rabbit hole in the first place), so this may not be right. Thought I’d share.

 ![fedora-weekly-actives](https://us1.discourse-cdn.com/fedoraproject/original/3X/2/9/296af3612a61a49be9f2f5e7dc8ef667b2f53217.png)

---

<div class="post-metadata">

**Author:** ![siosm](https://sea1.discourse-cdn.com/fedoraproject/user_avatar/discussion.fedoraproject.org/siosm/32/79_2.png) [@siosm](https://discussion.fedoraproject.org/u/siosm)\
**Post date:** [May 13, 2025, 7:53am UTC](https://discussion.fedoraproject.org/t/fedora-usage-stats/144635/11 "2025-05-13T07:53:51Z")

</div>

You should also include CoreOS in your stats 🙂

---

<div class="post-metadata">

**Author:** ![powerhouse](https://avatars.discourse-cdn.com/v4/letter/p/294172/32.png) [@powerhouse](https://discussion.fedoraproject.org/u/powerhouse)\
**Post date:** [May 13, 2025, 10:55am UTC](https://discussion.fedoraproject.org/t/fedora-usage-stats/144635/12 "2025-05-13T10:55:55Z")

</div>

Would be interesting to see how this compares to other distros.

---

<div class="post-metadata">

**Author:** ![nikodunk](https://sea1.discourse-cdn.com/fedoraproject/user_avatar/discussion.fedoraproject.org/nikodunk/32/4712_2.png) [@nikodunk](https://discussion.fedoraproject.org/u/nikodunk)\
**Post date:** [May 13, 2025, 3:16pm UTC](https://discussion.fedoraproject.org/t/fedora-usage-stats/144635/13 "2025-05-13T15:16:39Z")

</div>

> [@powerhouse](#):
>
> Would be interesting to see how this compares to other distros.

See [Fedora usage stats - #8 by popey](https://discussion.fedoraproject.org/t/fedora-usage-stats/144635/8) for that!

---

<div class="post-metadata">

**Author:** ![nikodunk](https://sea1.discourse-cdn.com/fedoraproject/user_avatar/discussion.fedoraproject.org/nikodunk/32/4712_2.png) [@nikodunk](https://discussion.fedoraproject.org/u/nikodunk)\
**Post date:** [May 13, 2025, 3:21pm UTC](https://discussion.fedoraproject.org/t/fedora-usage-stats/144635/14 "2025-05-13T15:21:30Z")

</div>

> [@siosm](#):
>
> You should also include CoreOS in your stats 🙂

With the same weekly active checkins [script](https://github.com/ublue-os/countme) @siosm 🙂

 ![image](https://us1.discourse-cdn.com/fedoraproject/original/3X/f/5/f5d73b036e1ff9aa174edbc4260740addab86560.png)

PS look at those spikes around workstation release date from all the hoppers 😃

---

<div class="post-metadata">

**Author:** ![jspaleta](https://sea1.discourse-cdn.com/fedoraproject/user_avatar/discussion.fedoraproject.org/jspaleta/32/54431_2.png) [@jspaleta](https://discussion.fedoraproject.org/u/jspaleta)\
**Post date:** [May 13, 2025, 5:28pm UTC](https://discussion.fedoraproject.org/t/fedora-usage-stats/144635/15 "2025-05-13T17:28:57Z")

</div>

In the very misty past, I had actually spun up a geolocated heatmap that ran somewhere in the infrastructure using geoip and matplotlib’s basemap package to “see” where the fedora machines were checking in from.

I think one release early on we ran it hourly and I did a animated gif so we could see people waking up on release day and doing the installs/updates roll across the map from east to west…

I was very proud of the fact that I personally made a pixel on that map light up in Alaska. Sadly the fedora system I installed in antartica get geoip’d to Denver. It would have been cool to light up the pixel for McMurdo Station.

---

<div class="post-metadata">

**Author:** ![nikodunk](https://sea1.discourse-cdn.com/fedoraproject/user_avatar/discussion.fedoraproject.org/nikodunk/32/4712_2.png) [@nikodunk](https://discussion.fedoraproject.org/u/nikodunk)\
**Post date:** [May 13, 2025, 9:55pm UTC](https://discussion.fedoraproject.org/t/fedora-usage-stats/144635/16 "2025-05-13T21:55:34Z")

</div>

- @jspaleta @siosm @mattdm @gwmngilfen if useful, i just merged a growth\_upstream.svg generator into [GitHub - ublue-os/countme: countme](https://github.com/ublue-os/countme) so if you clone and run that locally as in the readme, you can generate the chart below in a few seconds.
- You can also generate it for the past 5 years if you update the `months` in the script (also below).
- I also fixed a KDE Plasma edition mis-count - it’s way higher I think.  
KDE Plasma and CoreOS seem to be growing a lot!

 ![image](https://us1.discourse-cdn.com/fedoraproject/original/3X/8/c/8cbbab8bb3f95f661ab18cec9740fa333ce1be43.png)  
 ![image](https://us1.discourse-cdn.com/fedoraproject/original/3X/3/a/3a2fb923d2a951f76f9a8a07de63aecafc790306.png)

---

<div class="post-metadata">

**Author:** ![boniboyblue](https://sea1.discourse-cdn.com/fedoraproject/user_avatar/discussion.fedoraproject.org/boniboyblue/32/43330_2.png) [@boniboyblue](https://discussion.fedoraproject.org/u/boniboyblue)\
**Post date:** [October 18, 2025, 9:45am UTC](https://discussion.fedoraproject.org/t/fedora-usage-stats/144635/17 "2025-10-18T09:45:03Z")

</div>

@nikodunk: ~~Do you happen to have up-to date stat usage - I’d be curious to see how KDE Plasma is getting on how after it’s first 6ish months as an edition.~~

So turns out there was a link above that I missed. For anyone else interested.

 ![20251017 Fedora countme](https://us1.discourse-cdn.com/fedoraproject/original/3X/2/a/2aed10d0036483d25c5cc36902d1c42ea6289847.png)

---

<div class="post-metadata">

**Author:** ![nikodunk](https://sea1.discourse-cdn.com/fedoraproject/user_avatar/discussion.fedoraproject.org/nikodunk/32/4712_2.png) [@nikodunk](https://discussion.fedoraproject.org/u/nikodunk)\
**Post date:** [October 19, 2025, 4:49pm UTC](https://discussion.fedoraproject.org/t/fedora-usage-stats/144635/18 "2025-10-19T16:49:26Z")

</div>

Yep - quite some growth huh! [Here’s the link](https://github.com/ublue-os/countme) again for posterity - the file above is called growth\_upstream.svg and is updated once a day (though countme data only updates once a week)

---

<div class="post-metadata">

**Author:** ![tragivictoria](https://sea1.discourse-cdn.com/fedoraproject/user_avatar/discussion.fedoraproject.org/tragivictoria/32/56102_2.png) [@tragivictoria](https://discussion.fedoraproject.org/u/tragivictoria)\
**Post date:** [October 19, 2025, 10:23pm UTC](https://discussion.fedoraproject.org/t/fedora-usage-stats/144635/19 "2025-10-19T22:23:42Z")

</div>

1. You have very cute kitty on your pfp
2. That blog was really fun to read 😃 thanks for sharing

---

<div class="post-metadata">

**Author:** ![tragivictoria](https://sea1.discourse-cdn.com/fedoraproject/user_avatar/discussion.fedoraproject.org/tragivictoria/32/56102_2.png) [@tragivictoria](https://discussion.fedoraproject.org/u/tragivictoria)\
**Post date:** [October 19, 2025, 10:39pm UTC](https://discussion.fedoraproject.org/t/fedora-usage-stats/144635/20 "2025-10-19T22:39:15Z")

</div>

Sorry for stupid question (i don’t know anything about statistics), but why are there such huge spikes in usage in the 2020-2025 chart? That seems weird.

[Next page](https://discussion.fedoraproject.org/t/fedora-usage-stats/144635.md?page=2)
