Analytics answers which pages were visited and for how long. It does not answer what the visitor was trying to achieve, and identical event sequences routinely mean opposite things.
Treating a session as observable events generated by hidden states splits the problem into three levels: a stable motivation the visitor arrives with, an intent that shifts as they browse, and the clicks themselves. Forward-Backward and Viterbi then reconstruct the most coherent explanation of the entire journey rather than judging each action alone — and the states are regularised so they keep a meaning a human can actually read.