Survey methods

How many survey responses you need

A sample size table and formula for surveys, what margin of error and confidence level mean, and how many people to invite.

Last updated

How many answers are enough? It depends on how many people you want to speak for and how exact the result has to be. For most surveys, the table below answers it.

The quick answer

Responses needed at a 95% confidence level, depending on the size of the group you’re asking about:

Group size±10% margin±5% margin±3% margin
100508092
25070152203
50081218341
1,00088278517
5,00095357880
10,00096370965
100,000+973851,068

Beyond a few thousand people, the number hardly grows. That’s why national surveys get by with about 1,000 respondents.

What the terms mean

  • Group size (population): everyone the result should describe, like all 500 employees or all customers. It’s not the number of people you invite.
  • Margin of error: how far the true value may be from your result. With ±5%, a result of 60% means the real figure is likely between 55% and 65%.
  • Confidence level: how sure you can be of that range. 95% is the standard: in 19 of 20 surveys, the true value lies inside the margin.

The formula

For a large group:

n = z² × p × (1 − p) ÷ e²

  • z is 1.96 for a 95% confidence level (1.645 for 90%, 2.576 for 99%).
  • p is the share you expect, as a decimal. Use 0.5 if you don’t know. It gives the largest, safest number.
  • e is the margin of error as a decimal, so 0.05 for ±5%.

With these values: 1.96² × 0.5 × 0.5 ÷ 0.05² = 384.16, rounded up to 385.

For a smaller group of N people, correct it:

n ÷ (1 + (n − 1) ÷ N)

For 500 people: 384.16 ÷ (1 + 383.16 ÷ 500) = 217.5, so you need 218 responses.

How many people to invite

Not everyone answers. Divide the responses you need by the response rate you expect:

Invitations = responses needed ÷ response rate

If you need 218 responses and expect 30% to answer, invite 727 people. Internal surveys often reach 30–50%; surveys sent to customers by email usually stay well below that.

When the number matters less

  • Asking everyone. If you survey your whole team of 12, you don’t have a sample. Just try to get everyone to answer.
  • Looking for ideas. For open questions, 20 to 30 answers usually surface the main themes.
  • A skewed group. 2,000 answers from your most active fans still don’t describe all customers. Who answers matters more than how many.

Get to your number with StrawPoll

Related articles