A big one for me was realizing that a matrix times a vector returns a weighted sum of the matrix columns, with the vector's elements as weights. It's not particularly profound but it has definitely clarified my intuition around a number of matrix problems.