If you are performing an experiment, the goal is usually to get a representative view of a result in a certain target or population. Because of practical reasons (too big, too expensive, taking too much time, ...) it is often difficult to work on the whole population. In this case, a sample is used. This is a selection of the population, chosen to reflect the best possible population.
It is very important to use a correct size for your sample. If it is too big, it can create unnecessary costs and delays. When your sample is too small, the results are not statistically significant and no reliable conclusion can be drawn.