Hacker News Reader

Comment by dpcan | Hacker News Reader

Is anyone using an open source statistics system that's as simple as simple analytics? Maybe even something that can be setup to run stats for multiple clients from our own VPS?

View on HN

A relatively simple log analyzer, the kind we used to use in before que Analytics craze is GoAccess: https://goaccess.io/

If you want something more complex, Analytics style, you could try Matomo (Used to be called Piwik): https://matomo.org/

Shameless plug, but PostHog is MIT licensed, open source product analytics [0]. You can use it for super simple pageview analytics and even give all of your clients their own team on one instance.

If you want something more advanced, we're comparable in features to Amplitude or Mixpanel too, but more engineering focussed.

[0] https://github.com/posthog/posthog

Freshlytics - https://github.com/sheshbabu/freshlytics

* Cookies are not used

* Personally identifiable information (PII) is not collected

* See the pageview in different dimensions like page urls, referrers, browsers etc

* Supports multiple projects

* Supports RBAC

Screenshots - https://github.com/sheshbabu/freshlytics/blob/master/docs/sc...

I really like the look of this. And I really like the response here regarding "blockers" [0].

Is there any documented way to let users opt-out sans ublock:o (or other)? Or configure it to respect "Do Not Track"?

[0] https://github.com/sheshbabu/freshlytics/issues/12

Thanks!

> Is there any documented way to let users opt-out sans ublock:o (or other)? Or configure it to respect "Do Not Track"?

I initially wanted to implement DNT but there's a lot of confusion about whether it's useful or not. I'm not sure if there's a standardized way for users to opt-out.

How does it track individual users without cookies?

Yes, it doesn't actually track individual visitors as we can't use cookies or any fingerprinting methods.

Looks useful, but it depends on PipelineDB, a PostgreSQL extension for streaming data. Unfortunately PipelineDB hasn't been updated since May 2019 [0] when they were acquired by Confluent [1]. The former PipelineDB team appears to be focused on Confluent's KSQL product [2]. There's an open source "ksqlDB" but it appears to depend on Kafka, so it's not a 1:1 replacement for PipelineDB[3].

[0] https://github.com/pipelinedb/pipelinedb

[1] https://www.confluent.io/blog/pipelinedb-team-joins-confluen...

[2] https://www.confluent.io/blog/confluent-cloud-ksql-as-a-serv...

[3] https://ksqldb.io/quickstart.html

Yes, I'm relying on the "continuous views" feature of PipelineDB which is like autorefreshing materialized views. I'm planning to swap PipelineDB with TimescaleDB in near future

Most of the heavy lifting is done by Postgres/PipelineDB with Node.js as a simple wrapper so it's both performant and consumes less resources.

http://docs.pipelinedb.com/continuous-views.html

https://docs.timescale.com/latest/using-timescaledb/continuo...

There is also https://plausible.io/

I personally currently don't use any analytics, but they claim to be simple and lightweight so I keep an eye on them in case of future use.

Shameless plug, I wanted to develop a simple tiny stats dashboard using firebase as a backend: https://github.com/Karalix/feu-analytics

However, while it does what it is intended to do, it needs further dev that I am too lazy to do.

I would say go with Countly [1]. It is both for mobile, web and desktop, unlike several products mentioned here.

[1] https://github.com/countly/countly-server

GoatCounter[1] was featured on HN at some point. Open source, written in Go. It has a "subsite" feature which I haven't played with.

[1] https://www.goatcounter.com/

Reply on news.ycombinator.com