What? Why?
3 / 3.9 = 0.769 -> XB1 sell-through for November and December is 76.9%
Then assuming the same sell-through for January to March, we have
5 * 0.769 = 3.846 million units
It is a far far better approach than guessing a random number.
No it is not. I am all for extrapolating but only where it makes sense. # of units sitting on retailer shelves is not at all a percentage function of numbers shipped. It just isn't, not at all, not even close.