B2B Marketing Blog

How much is enough?

Aug 20, 2009 8:04:11 PM / by The Mezzanine Group

Here's a question that lots of marketing and business executives (who aren't research PhDs) face when they want to do market research: how many people should we survey in order to get a valid sample? It's a great question because it means people are thinking about doing valid research! Without going into the heavy statistics equations that answer that question, here are some guidelines on what sample size you need to gather statistically significant results:

  • The smaller your population, the larger percentage you'll have to survey in order to have a robust sample (+/- 5% margin, 95% confidence - ie, a 'statistically significant' sample). For example, for a population of 100, you'll have to survey 80 people (80%). But for a population of a million, you can survey 390 (.039%) and get the same statistical significance.
  • For a population of 500; survey 200
  • For a population of 1,000; survey 280
  • For a population of 5,000; survey 360
  • For a population of 10,000; survey 370
  • For a population of 100,000; survey 390

The only time you need more than 390 participants is when you have different segments in your population. That's why you see political polls with thousands of people, because they need statistical significance within each of the segments (for example citizens of each state / province; of various age groups, etc).

Even if you have a very small population, 30 is considered the minimum participation to do any kind of analysis.

Good luck with your surveys!

Topics: Marketing Consulting