Traditional book retailers pay suppliers 90 days after the book enters inventory whereas Amazon averaged about 58 days. The problem for traditional retailers is they held books in inventory (i.e. the book went unsold) for an average of 167 days versus Amazon's 16 days. This resulted in retailers carrying the cost of the book for ~78 days while Amazon was able to hold the float for ~41 days.
The end result of that type of inversion is that Amazon can accept a much lower margin, earn the float on the cash and live off much faster inventory turns than a traditional retailer. This was much more brilliant than "disintermediation" - as another poster has correctly noted, Amazon was an aggregator/replacement, not a true disintermediator.