Battle of the simple, privacy-focused analytics products | Hacker News Reader

Battle of the simple, privacy-focused analytics products

usefathom.com

3 pointspauljarvis13 comments

The competition being (non-exhaustive list, comment if you know of others I might have missed):

SimpleAnalytics.com, Plausible.io, Matomo.org, GoatCounter.com, and (shameless plug but WIP) Chiffre.io.

However, while there is a competition to gain users and market share, ultimately these platforms are in competition against Google Analytics and other giants of the surveillance industry. The only risk if one of them grows too fast is that it becomes the new giant. Google used to say "don't be evil", after all.

I think self-hosting is the solution. This way, there is no company that will store centralized user data, which is the biggest privacy issue for analytics.

I have been building a self-hosted platform [0], and I always get pressured into offering it as a SaaS. I am strongly against it and try to make it work by teaching people that self-hosting is really easy and it's the way of the future.

[0]: https://www.usertrack.net

> I think self-hosting is the solution. This way, there is no company that will store centralized user data, which is the biggest privacy issue for analytics.

Your opinion will change in time. It all comes down to the principles of the company and who has control of it. That's why Fathom hasn't taken venture capital. As soon as you bring in people who don't share the same values as you, things start going downhill. You do not need to store centralized user data to run a sustainable competitor to Google Analytics

I am slightly confused by your comment. Why do you say my opiniong will change in time? It has been the same over 8 years now and it just seems to get stronger and stronger.

"You do not need to store centralized user data to run a sustainable competitor to Google Analytics." That was exactly my point, right? Any platform that is hosted stores centralized user data (as they control all the servers), so self-hosting is the only way to decentralize the data. Am I missing anything?

You replied to a comment talking about:

> Fathom, SimpleAnalytics.com, Plausible.io, Matomo.org, GoatCounter.com, and (shameless plug but WIP) Chiffre.io

and you said :

> I think self-hosting is the solution. This way, there is no company that will store centralized user data, which is the biggest privacy issue for analytics.".

Your premise was that these companies store centralized user data. That is absolutely not true. So you presented a fake problem (that doesn't exist amongst privacy-focused analytics solutions) and gave a solution.

The company has access to all the data, thus it's centralized where one entity has full control over it. Can you further elaborate how those companies don't have access to their data? In case of an aquisition, how safe is all the data stored on all the servers that the company controls? What exactly is not true?

You shouldn’t keep centralized user data. I’m done here as it’s taking us both too much energy to explain our points. Because I run Fathom, I know about the data I keep, and there’s nothing user specific. Even in our queue, we never store user data...

I am not saying that the companies themselves are evil or hurt privacy, I fully support thenm. I just saying that by giving one entity access to the data of millions of people and websites you might run into privacy issues at some point, even if the company does everything to avoid that.

I agree that giving someone else your data is the issue. But as an alternative to self-hosting, I'm working on an end-to-end encrypted approach [1]. I can offer a service (much easier to setup and maintain than self-hosting), but I don't have the data (not in a useable/leakable way).

https://chiffre.io

One problem with hosted analytics that keeps getting more obvious recently is that adblockers automatically block the domain names of those trackers. I think hosted might still work, but you would have to offer custom domain names or find a way to avoid adblockers.

Also, what does end-to-end encryption mean? HTTPS already does that, right? If there's some extra encryption on top, does it mean that the client will encrypt the analytics data before sending it?

That sounds interesting, but it feels like the whole purpose is to hide the data from your company. How can you make sure that you don't have access in any way to the data, if you also store the user accounts that have access to this data? If you wanted, what would stop you from logging in as that user and viewing his data?

E2EE means the analytics data is encrypted by the visitor, using public key cryptography. The secret key to decrypt that data is only known of the website owner. Anything I store is encrypted, including that secret key, and the source key is derived from the user's password, which never hits the server (using the SRP protocol for authentication).

The only thing I can see is how much data a website generated, but not what this data means.

And yes, the whole purpose is to hide the data from me. I don't want to know who visits your website, how long they stayed or what pages they visited. The website owner may want to know, but I can tell them without having to know myself.

That sounds pretty cool.

I still think the company could somehow get access to the user data. The server enryption sounds pretty solid, and I beleive you saying that it would be really hard or impossible to get the encrypted data from the servers, but the information shown in the browser is unecrypted and accessible. Let's say your company grows and gets acquired by someone like Google and suddenly they want to get the stats of the tracked websites. Isn't there a way they could modify the UI source code, so that for example, they send back the unecrypted stats from the browser to a different server? I assume the dashboard's UI is hosted and served by your platform, so they could just add some JavaScript to it to send all data visible on screen, right?

I think your product could be really useful and I'm not trying to attack it in any way, I am just making sure I understand what the encryption means and if it does indeed provide extra value.

In theory, yes. But I consider Google an enemy of privacy, and I would rather burn my servers to the ground than sell them my product.

The encryption provides some protection for me: the data I acquire does not accumulate value over time (unlike clear-text data), which does not paint a giant target on my back.

Reply on news.ycombinator.com