Similar to the non uniform focal spot size in the beam the intensity distribution is also not uniform in the beam.
Most dose is near the center of the beam and to the edges the dose is reduced. Here is an example of a 20° tube (uncorrected gain image of a detector):

This effect could be reduced by the distance between tube and detector - but for the price of a loss in dose (double distance == quarter of dose).
Using a DDA this intensity distribution shall be corrected by the gain calibration (beside the pixel non uniformity). But also then there is a drawback: Due to the lower dose in the corners the SNR is reduced - in the image of the example by factor of 1.5. If a hole type IQI would be in the center with an CNR of 3.0 and another one near the corner, this one would reach a CNR of 2.0 only - which is less than the mostly required CNR of 2.5.