Opensource monitoring

I host my blog on Vercel and deploy via GitHub Actions from a private repo. I do plan on opening up the repo to the public evenutally but for now I'm rebuilding the Static Site Generator (rebloggy) that I've written in rust - it's a great learning project more than anything else.

I used to get some sense of traffic to the blog via the Vercel mangement pages and the rough number of requests to the various parts of the page e.g the CSS, HTML and images. However it seems that Vercel have changed this up recently and rather than adapting my workflow I figured I would use this as a chance to check out goatcounter.

GoatCounter

I first heard about goat counter by chance as I was reading up on open source self-hosted software that I could include in my raspberry pi cluster. Goat counter claims to be "meaninfgul privacy-friendly web analytics" and you can check out the source code over on their GitHub. I opted for the hosted version as I'm not quite ready to open my cluster to the internet just yet.

Screenshot of the Goat Counter homepage

The main stats provided are pretty simple and you can get a sense of the various visits to your site / pages as well as some other details that can be inferred from the browser e.g. language, client info etc. While lacking the robust features of Google Analytics I am pretty encouraged by the lack of any major off-platform tracking, no cookies or browser storage is used - simply a small blob of JS (which you can inspect for yourself here).

Why are you spying on me?

I find it super useful to be able to see the traffic my posts receive, especially ones from years ago like HomeKit â…¹ Roborock S5 Vacuum setup which still gets a reasonable amount of hits. As for the data collected, any session / user data is completely anonymised and I can't see any IP addresses or other ways to uniquely identify users.

Ultimately this is the sort of data I could collect from the HTTP Server if I managed the hosting myself, something simple like scraping the Nginx access.log would be enough to get similar visibility to GoatCounter. However I decided against self-hosting as I wanted to reduce friction to writing and having a simpler GitHub repo and automatic deployment was the way to go (if you know of some free metal-as-a-service that I can use to throw some files up then hit me up).

I don't think I will use this data as any more than just a nice reassurance that, occassionally, people do stop by, and people stop by more often when I post about it on socials.

Will I self-host?

Yes! Probably in the future, GoatCounter offers a pretty reasonable free tier of 2.5M hits which would last me many years, but I would like to self-host, if only for the fun of it and I expect the privacy considerations differ very little given the sort of data that's being collected.

First appeared on Trusty Interior, last update 14 Jul 2024