How the estimate works
The calculator multiplies three things you already know roughly: how many repetitive asks land on the team in a week, how long each one really takes once you count the context switch, and what an hour of the answering person's time costs. It deliberately ignores the second-order costs (the queue while they are on holiday, the deal that waited on a policy answer) so the number you get is a floor, not a pitch.
Why interruptions cost more than minutes
The research on task switching is consistent: recovering focus after an interruption takes far longer than the interruption itself. A two-minute answer in the middle of deep work is not a two-minute cost. That is why the same question answered by an AI employee is cheaper even when the human answer would have been quick: the human never leaves what they were doing.
What to do with the number
If it is small, close the tab; you do not have this problem yet. If it is a salary fraction, look at where the asks concentrate: one channel, one person, one topic. That concentration is what a Knowledge Manager absorbs first, and it is why teams start with one employee in one channel rather than an everything bot.