How big a sample?

Margin of error ≈ 1 ÷ √n.

"68% of marketers say…" From how many? Surveys and customer research are central to content and positioning work. Knowing roughly how much to trust a sample stops you overclaiming, and helps you call out others who do.

The move

For a yes/no style survey, the margin of error ≈ 1 ÷ √(sample size). 100 people → ±10%. 400 → ±5%. 10,000 → ±1%. To halve the error, you need 4× the people.

Worked example

Survey of 625 people

  1. √625 = 25
  2. 1 ÷ 25 = 0.04
  3. ≈ ±4%
  • This is the 95%-confidence rule of thumb, a slightly generous version of the textbook formula.
  • It assumes a random sample. 1,000 people from your own newsletter is not a random sample of "marketers".