Hey folks
Don’t know if this is the right place to post this, so apologies if not.
I’ve been playing around with an idea in my head for some time now, and sort of archived it for a bit, but it popped back into my head today, and now I gotta ask for some feedback on it.
In a nutshell it’s fairly simple. I was thinking about making server side web analytics based on logger data. Think of it as a custom Logger backend that will generate real-time accurate privacy friendly analytics data for you. Kinda like Webalizer (if anyone remembers that), that basically worked off off access logs.
Is this something you would use in your Phoenix projects, as an alternative to existing analytics solutions?
Above is some prior art in elixir. I’d be curious how yours would compare/contrast? Also what do you think about data rollups? Like, how do you go from storing events hourly to daily to yearly? Would you do something with postgres or click house?
Hi @gdub01
Thanks for the feedback. I was thinking about “streaming”/batch inserting the data into a BigQuery cluster with some multi tenancy, but that could very well be something different. I’d like that “reports” based on the analytics data could be built from SQL.
I didn’t know about the Dashbit article, I’ll check that out, thanks!.
We already use Plausible at work, and it’s fine, but it does not give you the full picture, because it’s (rightfully) respecting do-not-track settings, adblockers etc for privacy, plus it’s client side based.
I think that I can achieve the same level of privacy, but with a higher degree of accuracy with this
Indeed. The underlying architecture is very much inspired by Logflare.
I did however look a bit into ClickHouse, and it seems sick, and worth looking into I think
