- I'm going to capture 5% of users sessions as they progress through a purchase funnel
- I capture data for a few days after a design change, and look at behavioral changes, such as spending more time in a particular step of the funnel
- Where there's something weird going on, I'll watch an aggregate view of people's mouse behaviors - I can then see that a lot of people are hovering between two different buttons, or highlighting a particular sentence
- Based on that feedback, I'll do an A/B test of changing the button textx or the dialogue message
- Capture more data, did that change?
It's not "Let's go look at john smith's website behavior" except in the support example, where you get a user's consent to do that, and you usually only do it for internal corporate applications where there's a dedicated helpdesk.
Google does this on all of their products. And people that have left Google have made their own versions of this software for others to use. For example: https://www.fullstory.com/
I'm genuinely curious what steps are executed once this data is collected.
Also, in theory, you only need to do this after design/layout changes and only for a while, so what justifies collecting it all the time? I might be wrong though.