You are investigating if arrivals at a queue to use an ATM machine are well described by a Poisson distribution.
If you got to an ATM and there were already 4 people waiting to use it what might you do? How will this behaviour effect whether or not the Poisson is a good model, what assumption might be violated.