Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

Thanks! Similar to how Plausible is doing. Each user gets a unique identifier which is based on `hash(domain+useragent+ip+salt)` (the salt is generated each day).


That does create clusters for users that work at the same company, with the same company-issued devices. Or iPhones in a New York City block.


A bit more factors tbh. User agents have browser, version etc.

iPhone in New York block still have different IP, versions etc

But I hear you, it’s better than cookies and that’s the entire point.

You can also provide a user id, more info here https://news.ycombinator.com/item?id=40434115


This is the same problem that I had with parsing logs. There's no way to know that mom+dad+kid all using the same model device were 3 different users.


If the salt changes daily, how is. It possible to know how many users returned after > 24 hrs?

Edit: @lindesvard- Got it, thanks for clarifying!


You'll never know any returning users after salt changes. This is the hole point with privacy first.

But you have options tho. Compared to other tools that is privacy first, you can actually pass your own user id. You can get this ID from where ever you want (db, cookie, localStorage. Its up to you whom implement it)

But I want to have a middle ground where privacy is important but you have the options to decide what path you take.


So maybe an obvious question but just to be clear, if you pass your own user ID you can track that particular user's flow across multiple sessions/days/etc?


Yes! So its a matter of priorities. Is privacy or retention most important? Than you can decide how to track users.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: