Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!


Alternatives to Google analytics
New on LowEndTalk? Please Register and read our Community Rules.

All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.

Alternatives to Google analytics

What do you all use?

I tried goatcounter and it isn't compatible with cyberpanel. Probably reverse proxy is needed.

«1

Comments

  • I use my nginx access logs

  • yoursunnyyoursunny Member, IPv6 Advocate
  • Flattering

  • Paid? There are some.

    Fanthom Analytics
    Fanthom Analytics

    or Matomo Analytics
    Matomo Analytics

    Thanked by 1ariq01
  • Free and Open-Source: https://umami.is/

  • Daniel15Daniel15 Veteran
    edited December 2023

    I'm self-hosting Plausible and it's working well enough for me. I tested a few others, and it was the best option. https://plausible.io/

    Their tracking script is <1KB and it can be bundled directly into your app's JS (via their npm package) instead of having to load it separately.

    It's cookie-less, and user session IDs are derived from a hash of the IP address and user-agent, which means it's GDPR- and CCPA-compliant. It also lets you import all your old Google Analytics data.

    It also uses a column-oriented database (Clickhouse) which makes it extremely efficient - aggregating 5 years worth of data is no problem. Column-oriented databases are optimal for data where you mostly aggregate columns (e.g. get total number of hits in the past year, get average visit duration over the past month, etc). If you want to self-host, look for something that uses Clickhouse.

    It has the same downside that all cookieless trackers have: Unique visitors are undercounted. If multiple people access your site with the same browser from the same IP, they'll just be seen as one unique user. The same with ISPs that use CGNAT - every person with the same browser and the same public IPv4 will be seen as the same person. All other metrics should be OK though.

    It's pretty basic though. If you want something more fully-featured, Matomo is a good option, but it's huge and inefficient. Queries are far slower than something that uses Clickhouse (like Plausible) and it uses more storage space.

  • Posthog: https://posthog.com/

    Self host, or it has a generous free plan.

    Thanked by 1sillycat
  • https://matomo.org/ has a free plan.

    Thanked by 1jfreak53
  • tentortentor Member, Patron Provider

    @Moopah said:
    I use my nginx access logs

    Mentally strong website owner has data science degree and is proficient in python for data analysis

  • ailiceailice Member
    edited December 2023

    I using GoatCounter (https://www.goatcounter.com) on my website and its simply amazing
    EDIT: Looks like you aleardy mention goatcounter perhaps test matomo?

  • My website uses Cloudflare Web Analytics.
    However, I don't pay much attention to references from search engines these days.

  • Thanked by 1sillycat
  • I self host https://umami.is/ and it's awesome. It's similar to Plausible Analytics (which can also be self hosted) but uses regular Postgres instead of Clickhouse so it's easier to maintain and back up.

    Thanked by 1ariq01
  • I am self hosting Umami as well. Frontend on Vercel and using PlanetScale for database. I have it running on >10 websites with 4 separate Umami users.

    Some other options you can try:

  • Matomo (self hosted), Matomo free tier ( matomo.org), or , if you are using wordpress, you can use the matomo wp plugin.

  • Daniel15Daniel15 Veteran
    edited December 2023

    @vitobotta said: uses regular Postgres instead of Clickhouse so it's easier to maintain and back up.

    Clickhouse isn't too difficult to back up, and it's got support for incremental backups to disk or S3 built-in.

    Querying it is also going to be much faster than Postgres, especially for queries over longer time periods. For example, I imported 10 years of Google Analytics data for three sites into Plausible (one of which gets 150k+ hits per month), and viewing all the data still takes less than 100ms to load, and the DB only uses 300MB on disk.

    Row-oriented databases like Postgres and MySQL are good for loading particular rows or small aggregations, but they can get very slow if you want to load a lot of data from one column.

    Column-oriented databases like Clickhouse are the opposite - they're good for querying individual columns even over very long periods of time (eg get the average number of page visits per unique visitor over the past three years) but not good for loading particular rows. Analytics is all about aggregation of columns so they work very well in that case.

    It's all about using the right tool for the job :D

  • @Daniel15 said:

    @vitobotta said: uses regular Postgres instead of Clickhouse so it's easier to maintain and back up.

    Clickhouse isn't too difficult to back up, and it's got support for incremental backups to disk or S3 built-in.

    Querying it is also going to be much faster than Postgres, especially for queries over longer time periods. For example, I imported 10 years of Google Analytics data for three sites into Plausible (one of which gets 150k+ hits per month), and viewing all the data still takes less than 100ms to load, and the DB only uses 300MB on disk.

    Row-oriented databases like Postgres and MySQL are good for loading particular rows or small aggregations, but they can get very slow if you want to load a lot of data from one column.

    Column-oriented databases like Clickhouse are the opposite - they're good for querying individual columns even over very long periods of time (eg get the average number of page visits per unique visitor over the past three years) but not good for loading particular rows. Analytics is all about aggregation of columns so they work very well in that case.

    It's all about using the right tool for the job :D

    No doubt Clickhouse is better suited for analytics at larger scale, but for small to medium sized sites Postgres is fine. Clickhouse was a pain to back up was I was self hosting Plausible because some tables were not in a format supported by the Clickhouse backup tool, but can't remember the details. It's possible that his has improved in the meantime. Still, I like the simplicity of just using PG with Umami, because I don't have millions of visits per day on my sites.

  • varwwwvarwww Member
    edited December 2023

    I am using Microsoft Clarity (https://clarity.microsoft.com/) currently along with Google Analytics. It's pretty epic and it is free. Has heatmaps and session recordings (realtime).

    I also use

    • goaccess https://goaccess.io/ (for raw server data analytics - access.log*) This is required since most of other analytics scripts will get blocked by ad blockers.
    • Cloudflare dns analytics (from the dashboard)
  • Daniel15Daniel15 Veteran
    edited December 2023

    @varwww said: This is required since most of other analytics scripts will get blocked by ad blockers.

    Plausible won't, since you can embed it directly into your site's JS or load it from your site's domain (meaning no third-party scripts to load), and you can also proxy the logging endpoint so all log calls are same-origin requests (requests to the site's domain). That pretty much guarantees it won't be blocked by an adblocker.

    Server-side analytics are good, but an issue with them is that you'll see a LOT of traffic from bots, since it's practically impossible to differentiate bots from real users just from the small amount of data in server-side logs.

    Thanked by 1varwww
  • varwwwvarwww Member
    edited December 2023

    @Daniel15 said:

    @varwww said: This is required since most of other analytics scripts will get blocked by ad blockers.

    Plausible won't, since you can embed it directly into your site's JS or load it from your site's domain (meaning no third-party scripts to load), and you can also proxy the logging endpoint so all log calls are same-origin requests (requests to the site's domain). That pretty much guarantees it won't be blocked by an adblocker.

    Server-side analytics are good, but an issue with them is that you'll see a LOT of traffic from bots, since it's practically impossible to differentiate bots from real users just from the small amount of data in server-side logs.

    Heard of Plausible before but did not know it was open source. Thanks.

    Bot traffic is easily removable from the server side logs. I have done it myself. Just identify the pattern - remove GPT bot traffic, search engine traffic, SEO marketing bot traffic, curl/python-/go-http traffic and the remainder is usually mostly human traffic. Yea, some of them will have spoofed user agent but it is far more reliable than all the analytics tools out there.

  • jfreak53jfreak53 Member, Patron Provider

    Matomo, been using it since it was Piwik v1. Open source, amazing, and I own my data.

  • @jfreak53 said:
    Matomo, been using it since it was Piwik v1. Open source, amazing, and I own my data.

    Seconded. Never used Matomo but Piwik was pretty nice a couple years back.

  • Adding to this thread: there's a short list of open-source analytics tools on awesome-selfhosted I came across the other day while poking around for answers to this same question.

    Personally, bounce between Plausible and Umami—there's features of both that I like.

  • crunchbitscrunchbits Member, Patron Provider, Top Host

    Matomo is solid. Plausible was cleaner, but needed some extended functionality.

    IMO, we need to go back.

  • @crunchbits Don't forget to set the initial value to a few thousand so it looks like you're more popular than you are ;)

    Thanked by 1crunchbits
  • I use Microsoft Clarity because its fairly light, has heatmaps and session recordings. Panel is great. Its 100% free, but ofc data is being used to improve other Microsoft services such as Bing.

  • @AXYZE said: 100% free

    @AXYZE said: data is being used to improve other Microsoft services such as Bing

    Thanked by 1COLBYLICIOUS
  • UchihaUchiha Member
    edited December 2023
  • @crunchbits said:
    IMO, we need to go back.

    I'm also surprised that everyone posts options like "you need Kubernetes with Node.js built-in running inside a Docker container to have proper Clickhouse support, which is required for our self-hosted stats solution that will run happily on 2 cores and only 4 GB RAM (for 1-4 users)".

    And no one still mentioned good old StatCounter.

Sign In or Register to comment.