While Binomial r.v. counts number of "success" in a fixed number of trial, the assumption is the trials are independent and have same probability for success. As long as an example approximately falls into this setting, we can use the distribution. Examples:
Consider the population of all registered voters in Islamabad, NA-19. Suppose this is 200,000. Let R1, R2, ..., R500 be a sequence of simple random draws from this population. This is a simple random sample of 500 Registered Voters. Surveyor ask each person "Are you planning to vote in the coming elections?" If voter responds "YES" we record 1, If voter responds NO, we record 0. Suppose that 150,000 people are planning to vote, and will respond YES (ignore the possibility that people might LIE to the surveyor). Then for each random draw, the probability of getting a YES is 150,000/200,000 = 75%. So each random draw generates a Bernoulli RV with success probability p=75%. Let S be the total number of people who say YES. Then S is a Binomial Random Variable: Bi(500,75%).
Remember to keep the PRE-EXPERIMENT and POST-EXPERIMENT concepts clear. S is a random variable when we are PLANNING to carry out the experiment (of drawing a random sample of 500 people and asking them whether or not they will vote). AFTER the survey is done, 500 registered voters have been chosen and have been asked the survey question. NOW we will have some number, such as 400 people said YES and 100 people said NO. 400 is the number of SUCCESSES that occurred. But 400 is NOT the random variable S. The random variable was DESTROYED when the experiment was carried out -- no more randomness left. 400 is the OUTCOME of the experiment. It is a fixed number and NOT a random variable.