Can neural networks be used to classify users?

Question

Project management

The fintech web service has tens of thousands of active registered users per day.
A significant part of them use one product within the service (several products).
But some at some point switch to using two or more.
There is a desire to cluster them by type of behavior in order to reveal the "path" from the state "uses 1 product" to the state "uses 2+ products".

Answer 1

It is possible, but IMHO it is "from a cannon to sparrows." It is better to use some kind of tags that are attached to the user and a suitable product is selected based on these tags ... well, at least I would do that.
Although neural networks can give an estimate of the approximate user conversion for a particular product. Basically, you need to experiment.

Answer 2

Log:
1. Client data: Acc, headers, session, etc.
2. Transitions between pages (URL chains)
3. Behavior on the page (hovers and focuses of elements, mouse coordinates, clicks, page position in the ViewPort (scroll), keyboard input, tab focus)
Then save the whole thing in the database and then mine any data.