Discrete data randomness

#3
So do you expect 50% to be "Cancelled" and 50% "Completed"?
Let me explain:

My average monthly jobs are 70k,out of which 20% are cancelled and 80% are completed.

Now we do random survey of approx 3000 customer in a month to gauge NPS and stakeholders at times not always raise a question,if our data for survey was random or not.

one way is to show simple % in overall data vs % in 3000 survey data.but i want to know an effective way in statistics.