Tool
How many people do you need to survey? It depends on the margin of error you accept, not on the size of your market.
The formula starts from universe size, the margin of error you accept and the confidence level, and applies the finite population correction. Expected proportion is left at 50% when unknown: it is the most demanding scenario and therefore the safest.
Watch what happens when you move the universe: between 100,000 and 10 million people, the required sample barely changes. What really moves the number is the margin of error, because it enters squared: going from ±5% to ±2.5% does not double the sample, it quadruples it.
Tell us what decision you are facing and we will tell you which study answers it.