Assume I need to conduct a t-test. I have a treatment and control group. I go ahead and measure the total number of minutes spent on a website by every user every day for N days. So for every user in the treatment and control group, I know the number of minutes they have spent on a website every day for N days

We want to conduct a test on the difference in minutes spent browsing the website for both groups. When conducting this test, should an observation represent: tot minutes by each participant in each group (so # rows = # participants) OR the total mins per group per day (row = #days)? which one would violate assumption and why?