In preparing for ATmosphereConf 2027 I wanted to get some insights into the European atproto community: how big it is, PDSes they use, apps/lexicons they use.
So I started collecting that data for EU in general and also (since I’m Dutch and involved in organizing some local meetups) for The Netherlands as well.
… and also aggregate/roll these up to the EU level
So this escalated nicely into me reaching out to a lot of people working on local events (which was fun and still ongoing!) and building https://tangled.org/atcommons.eu/website/ . This repo is now serving 6 local country websites + the EU aggregate website and still in talks with a couple others (still looking for the owners of atproto.de, atproto.co.uk and atproto.dk if you happen to know them :))
(And no I’m not squatting all these domain names, I’ve just asked the existing owners to update their DNS and most did!)
As an aside, I’d like to mention that last year, a media metrics company in Norway included Bluesky in their user count survey, and to their surprise, they came out estimating that Bluesky had 300 000 Norwegian users. The survey was discussed in detail by the Norwegian Broadcasting Corporation: Hundretusener av nordmenn bruker Bluesky: Nesten halvparten så stort som X (I’ll let you choose your own translation tool). They insist their methodology is robust. So, if that’s correct, it is quite a lot more than what you hit.
Unfortunately, the sources aren’t open. They are a survey company, they regularly ask people things, and 7% indicated that they used Bluesky weekly. So, they might have hit a very biased group, but it is still very interesting.
Also strongly suspect the number has dropped precipitously since 2024, as Bluesky didn’t reach the critical mass needed to stick in Norway the way it did in the US.
Ah yeah, I’m not doing any surveys but trying to source the actual data and build a more data-based (but still educated guess) from there…
The reason it dwarfs our current 58K is that they measures a much wider circle: looks like they asked 1,253 people on their GallupPanelet web panel which social apps they use, then scaled that up to the whole adult population. So it’s a projection from a sample, quite different from counting accounts.
The 1253 is fine to be representative, but it’s…
a self-selecting panel (of people that are more online, more survey-willing, plausibly more likely to be early adopters of a buzzy new social app). Even for a society like Norway that is very online this skews the numbers.
they only got a ~3% response rate (1253 of ~40k invited) which adds nonresponse bias
this survey ran late December 2024, peak X-exodus press when Bluesky was in every headline.
All 3 will drive the eventual estimate up.
Having said that: where the above might have a bias up, it’s still useful a soft upper-ish figure, order-of-magnitude. The atproto.no numbers (and the ones for other countries) are more conservative, so the truth probably lies somewhere in between. So even if we can’t treat the 310K as a solid point, the directional input seems solid and is useful
So I can’t really fold the 310k into our own count (it’s just not the same unit of what we are measuring). But still a great external validation to show alongside, I’ve add it to the .no stats page:
