We Rebuilt Fathom Analytics from the ground up and moved to Laravel Vapor | Hacker News Reader

We Rebuilt Fathom Analytics from the ground up and moved to Laravel Vapor

usefathom.com

35 pointsJackWritesCode26 comments

Open articleView on HN

I built a similar privacy focussed analytics inspired by Fathom:

https://github.com/sheshbabu/freshlytics

Privacy

* Cookies are not used

* Personally identifiable information (PII) is not collected

Dashboard

* Switch between multiple projects

* Slice the data within a date range

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

Tracking

* Currently focuses on website pageview tracking

* Tracks visited page urls, referrers, browsers and browser versions

Projects

* Supports multiple projects

* Different projects can have different reporting timezones

Users

* Users can either be Admin or normal users

* Admins can create/edit/delete projects and other users

Written using TypeScript, React, Postgres (PipelineDB), Express.

Here are some screenshots: https://github.com/sheshbabu/freshlytics/blob/master/docs/sc... :)

Ouch, this move and the plan to abandon the OSS part (discussed in the comments here) makes me uncomfortable about using Fathom.

The option to run my own instance if wanted was the main point for paying for their service. Because I don't want to maintain an instance of Fathom myself now, but maybe later. Now potentially giving up the OSS version makes me reconsider alternatives to Fathom.

All these changes lately make me feel bad about their future and their grand promises on my users' privacy.

Hey there,

Appreciate the comment. Good or bad, anything that makes us think about things is welcomed, so thank you!

1) The OSS version isn't going anywhere (listen to our podcast from 10 days ago, we talk about it: https://fathom.transistor.fm/episodes/the-open-source-dilemm...)

2) If you don't trust us, you shouldn't use our service.

3) Our fundamental guarantee regarding user data is that we don't sell it and we never will. We are already profitable and don't need to sell data.

I do want to say I've been following the Github repo for a while now, and every time I notice a new update I get very excited, then disappointed because it's just a one line documentation change, haha.

Giving it a listen now though, headphones just finished charging.

Awesome. And yeah, we haven't made many updates to the codebase, sorry! Getting Fathom V2 out the door was priority number 1, as we are not currently full time on Fathom. The plan is to grow Fathom to support us near full time so we can spend more time on things. I'd also like to learn Go. We both have full time jobs outside of this.

There will ALWAYS be an OS version. We don't know how the future will look but we know we have to keep a free privacy-focused analytics option available. It's the right thing to do.

I also really liked the open source version for side projects, maybe a community fork could exist to fill the gap?

It totally could!

I would suggest listening to https://fathom.transistor.fm/episodes/the-open-source-dilemm... too (published 10 days ago) as that'll give insight into our plans.

The biggest problem with the open source version is that it's written in Go and there aren't that many people with both the skills and the time to maintain it, let alone expand it. It might be easier for a group of people to get together and build out a privacy centric open source project from scratch (something I'd be interested in).

Fantastic idea! Don't wait for someone to start it, start it yourself!

We're still going to be updating the OS version of Fathom but I 100% support anybody creating another privacy-focused analytics platform. Additional alternatives to Google Analytics are great!

Good luck :)

Has the open source version that's actually usable on prem for privacy-conscious sites been completely shitcanned at this point? I haven't seen or heard any updates at all about it and every link to it has been removed from the site.

Hey there,

Thanks for the love! The OS version has not been shitcanned. We spoke about our plans on our podcast 10 days ago: https://fathom.transistor.fm/episodes/the-open-source-dilemm...

Running 2 software products & working a full time job is hard but we'll get some updates to the OS version. We're likely going to hire a contractor to help us!

At this point it's completely diverged from the SaaS version, and is not similar in any way, correct?

Are you porting the SaaS version? Would you be willing to do some kind of paid "license" to run the SaaS version locally?

Great questions.

1) They are different codebases. OS was written in Go by the previous developer. I rewrote everything in Laravel

2) We don't think we'll be porting it to OS but there's no decision on this yet

3) That's a great idea regarding a license, because a lot of people want to self-host... We'll have a think because we've purposely tied our app to certain infastructure, so we'd need to put in significant work to make it transportable. I'm not saying never, I just don't know.

We address some of the complications on the podcast.

These are solid questions, thank you.

Do you think with the new Laravel stuff, 2) might be more feasible? I write a surprising amount of PHP for fun (not work/profit, so not really critical either) and IME Laravel has been incredibly easy to port out, with native config file support, CLI migrations, etc.

I think 3) is interesting, because effectively a huge draw of your product is targeting privacy conscious people, and those are probably the least likely to want to trust a third party for it.

I used https://haveamint.com/ for a while, but it's also been dead for quite a while. I haven't been able to get Piwik to handle my traffic volume.

I don't make anything from my side projects, but I prefer self hosting to know exactly how much data is being collected and reduce it or modify the tracker script as needed, location-respect by hosting analytics data in the same datacentre as the site (so EU users aren't randomly getting analytics ticks sent to US or whatever, some people care, this is just easier for me), and it's really, really cheap to have bulk capacity for me compared to paying a SaaS per pageview - I have about 4 TB of RAM free right now on my (mostly k8s) clusters.

So it's not Laravel that's making it hard, it just means that we'd have to rethink how we've tied our product to the infastructure. It's definitely possible, and if we go down that route, the software in Laravel will make it a breeze to export. We spoke with a chap on twitter who made a great point that we'd likely see more contributions since it's in PHP / Laravel. Less people know Go, and one of the big reasons we didn't feel the obligation to OS the new version was because next to nobody contributed to the Lite version.

And you're right. There are a segment of people who won't use analytics unless they're self-hosted. Totally understandable. We'll see what happens. It's not on the roadmap in the immediate future but we're not ruling it out.

I'm that chap that spoke to them on twitter. I want to be very clear about what happened- one of the cofounders blocked me on twitter for asking these questions (even though I was polite about it), and then the other one reached out and we had a pretty decent call. Unfortunately that call made it very clear that they were not going to make the new version open source.

Until I spoke to you, I thought you were a troll and so did Paul :P That call with you was very useful and gave us a lot to think about. We recorded a podcast about OS and I mentioned our call: https://fathom.transistor.fm/episodes/the-open-source-dilemm...

I asked the same questions on Twitter. One of the cofounders had a meltdown over the questions, and the other founder actually hopped on a phone call with me to discuss their plans.

As far as I can tell the open source version (which hasn't had any updates other than readme changes since earlier this year) is basically on life support. They have no plans to make the new hosted version open source, and only vague plans on hire a contractor in the future to keep the open source version alive. According to the cofounder I spoke with they went back and force about this half a dozen times, which is why their messaging has been so mixed- they had originally said in previous HN and twitter threads that they would open source the new version, but went back on that because they were worried it would take money from their hosted version.

Pretty much bang on. The only piece I'd point out is that Paul didn't have a meltdown, he thought you were a troll.

We're looking to make the business sustainable so that we are in this for the long run. Without money, we couldn't spend the same amount of time on it that we are.

And we were definitely bad on the messaging and are going to be more careful in the future as it does cause upset.

I asked about the open source version, and then made a tweet about how you had changed your mind (which you've admitted to here, and on our phone call). Your cofounder showed up, accused me of lying, and then blocked me before I could respond with screenshots of your company previously committing to open sourcing the new version.

If anyone was a troll in that situation it wasn't me ;-)

You're not a troll, we thought you were. We were very wrong. You were the complete opposite and wanted to help.

Paul has publicly admitted that he was wrong on the podcast. It is funny in hindsight. He denied that we said we would OS it then I reminded him that he made the graphic announcing it!

Big mistakes made on the messaging of this and lessons have been learned.

Glad I could be helpful! If you ever decide to open source the new version feel free to reach out if you'd like an advisor (or even just someone to bounce ideas off of). Open Source Sustainability is a huge issue for me, and I'm always happy to help open source companies.

There’s some negativity in this comment section, so I just wanted to say that I really like your product.

I’m glad you’re working on making it sustainable before spending more time on the open source version.

Appreciate the nice comment. And we always expect to be challenged when we post to Hacker News. It keeps our minds sharp as people ask some good questions ;)

Curious... what your November costs vs September costs ended up being. Did vapor actually end up saving $$. It's probably a better user experience I'd imagine.

We're on track to hit around $450 this month. Includes a ton of migration-related spikes though. Comparatively, Heroku would've been $700-750.

Also, we have over-provisioned cache & RDS with Vapor, whereas Heroku we were running on the edge.

Reply on news.ycombinator.com