Step 2 in Market Research: Sample Size - how many people do I talk to?

Still staying on the subject of improving your business by first getting more info... Now we have established who we want to talk to (Target Audience), let's consider the next step - how many people do we need to survey?

Step 2. Consider numbers

Most clients start with 1,000 completed interviews in mind. If you think of a sample of blood no bigger than a teaspoon as containing enough data to tell doctors about the state of your health, then that's a reasonable analogy as to why 1,000 interviews should tell you enough about what the population is up to for you to be happy. In statistical terms, a sample of 1,000 will give you accuracy of ±3%. So if 500 respondents say yes and 500 say no, the 50:50 split could be 47:53 or vice-versa. But doubling the number of respondents to 2,000 will only improve the accuracy to ±2%. 3,000 respondents will give you ±1.8%. You could potentially spend a lot of extra money for a comparatively insignificant improvement in reliability. There may be reasons to talk to more or fewer people. If you need to do some segmentation - for example you want to interview a significant number of purchasers of 20 different brands - a standard sample of 1,000 may be a bit thin. Bear in mind that just because you want us to find 200 people who own a Ferrari GTC4Lusso it doesn't mean we can. And that is an extreme example of why you may want to target fewer than 1,000 interviews. When incidence is low, our expectations should follow suit. We'll always give an informed view of how realistic your objectives are, and even tell you if we think you shouldn't go ahead.

Step 3. See next post: Survey Questions