Is there any way to track unique users following a feed?

[jay-tuckey]

Thinking out loud - is there any way to track unique users following a feed?

It would be useful for sites to be able to figure out how many unique users are fetching a feed URL.

Currently the User Agent from the logs looks like this:

Apr 30 15:10:36 nt-gov-newsroom-rss run_prd.bash[7331]: [2024-04-30 15:10:36 +0930] [7331] [INFO] 10.1.1.9:50224 - - [30/Apr/2024:15:10:36 +0930] "GET /rss.xml 1.0" 200 2132197 "-" "Mozilla/5.0 (compatible; Miniflux/2.1.3; +https://miniflux.app)"

Actual user agent:

Mozilla/5.0 (compatible; Miniflux/2.1.3; +https://miniflux.app)

I see that each feed is fetched individually by each user account, from this discussion:
https://github.com/orgs/miniflux/discussions/1797

Other hosted feed readers add the number of subscribers when fetching the feed, eg NewsBlur's User Agent:

NewsBlur Page Fetcher - 7 subscribers - http://www.newsblur.com/site/3966817/analytics-piwik (Mozilla/5.0 (Macintosh; Intel Mac OS X 10_7_1) AppleWebKit/534.48.3 (KHTML, like Gecko) Version/5.1 Safari/534.48.3)

This would be a bit harder in miniflux considering a lot are self hosted, so you'd need to add the primary URL to the user agent, as well as the subscriber count, which probably isn't possible given the "one fetch per user" situation.

Maybe instead, a UUID could be generated for each user and included? Then the UA would be something like:

Mozilla/5.0 (compatible; Miniflux/2.1.3; +https://miniflux.app; +miniflux_user=fcfb0192-a062-46fa-bedb-2c9be5104fc5)

I think this would still be pretty privacy respecting as a randomly generated UUID wouldn't give away any PII for the user. Could even do a UUID/per-user/per-feed if there was concern about matching users between feeds?

Keen to hear other's thoughts on this. :)

[fguillot]

There are some privacy concerns about advertising the number of subscribers (or a unique user ID) in the user agent.

Technically, displaying the number of subscribers at the instance level is probably feasible, but it's not something that aligns with Miniflux's principles (privacy and no user tracking). Additionally, I'm fairly certain that most people who self-host Miniflux are the sole users of their instance.

[jay-tuckey]

Totally agreed, would need some good thought to get the privacy right, but I think it shouldn't reveal anything newly identifying to include a randomly generated user ID with each fetch? The only new info in my mind is the number of users behind a specific server ID?

Not sure that there's that much demand for this feature, though, but seems like it could be useful for sites to get a general feel for number of subscribers.