Choosing your ideal sample size

On the one hand, you might feel you want as many people’s views as possible. But you might also have heard that, beyond a certain point, more respondents doesn’t equal more robust insights. It’s actually possible to reach ‘peak robustness’.

It’s up to you to balance your need for reliable research with the practical requirements of your research project.To put it bluntly, the more respondents you get, the longer it’ll take (and the more money it’ll cost).

Choosing a robust sample size

Luckily, there are some standard formulae to think about when choosing your sample size.

The calculation takes into consideration three factors: 

• Confidence level
• Margin of error
• Survey type

This will then give you the minimum number of respondents needed to achieve the level of confidence you want. 

Adding to the sample size can increase your confidence in the data, and will deliver results you feel comfortable presenting to your team and partners.

A confidence level of 95% and a margin of error under 5% are the industry-wide accepted minimums for good consumer research.

As a simple guide, for data that provides a nationally representative sample of the UK’s working age population (roughly 45 million people), here are the sample sizes to bear in mind. 

3,500 responses 

• Confident level: 99%
• Margin of error: 2%

2,000 responses

• Confidence level: 95%
• Margin of error: 2%

1,500 responses

• Confidence level: 99% 
• Margin of error: 3%

900 responses

• Confidence level: 95%
• Margin of error: 3%

500 responses

• Confidence level: 95%
• Margin of error: 4%

The larger the sample size, the more confident you can be in your findings. But for fast decision-making, a smaller sample size often works just as well—and it’s still much better than simply asking around the office (or working from a hunch!)

Larger sample sizes also allow you to drill further down into the data, offering a greater level of granularity and confidence in each demographic.

To put this into numbers:

• Directional (e.g. for brainstorming innovation ideas): 250-500 responses
• Statistically robust (e.g. for choosing innovation projects to prioritize): 501-2,000 responses
• Major, strategically important decisions (e.g. for major product launches): 2,001-4,000

Think about how question type affects sample size

When gathering quantitative data, you’ll generally want a large sample size for better accuracy.

If you’re gathering qualitative data, you’re probably looking for an overview of opinions within your target audience. But keep in mind that free text answers take more time to analyze—you’ll need to physically read the answers and bundle terms together to be processed.

Pro tip 💡

“Sample size depends on two things—the audience and the type of research.

For brand tracking, we advise that the ideal sample size is 1,000, but 500 will still be robust. For creative testing, other studies, or for very niche audiences, sample size can be smaller.”

Nick White

Customer Research Lead

Increase sample sizes for published data

Data for internal decision-making purposes varies from that which will be used and distributed more widely (for instance, to the press).

Pro tip 💡

“We recommend using a minimum sample size of 2,000 if you’re publishing the data externally.”

Elliot Barnard

Customer Research Lead