How many survey responses does a B2B study need?

Consumer surveys frequently need thousands of responses. B2B surveys often need far fewer. This guide gives you typical sample sizes for published reports and brand trackers, plus a calculator to check any sample size you're considering.

Margin of error tells you how far your result could be from the truth.

Margin of error tells you how far your survey result could be from the truth. Say 62% of your 400 respondents tell you they plan to buy, and your margin of error is 5 points. You can read that result as “somewhere between 57% and 67%.” If you ran the same survey again with a different 400 people, you'd almost always land inside that range.

Before you act on a result, picture both ends of the range. If you'd make the same decision at 57% and at 67%, you can act with confidence. If you'd move forward at 67% but hold back at 57%, increase your sample size so the range narrows enough to land on one side of your decision.

B2B surveys need fewer people than consumer surveys.

Consumer researchers slice their results many ways. They often compare regions, age groups, income bands and household types, and every slice needs enough people in it to be meaningful. That's a key reason consumer studies have sample sizes in the thousands. B2B studies often focus on a few key comparisons, such as company size and job title, which can reach reliable results with a few hundred respondents.

Cost is the second reason B2B samples stay small. A senior decision maker is hard to find and those professionals are unlikely to spend time filling out a survey for minimal compensation, so each B2B response costs more than a consumer response. Most B2B budgets can't stretch to a thousand people, and they typically don't need to.

40 studies across 9 publications had a median sample size of 410.

We reviewed 40 surveys of business professionals published by The Wall Street Journal, Bloomberg, Bloomberg Law, Forbes, Fortune, Inc., The New York Times, Harvard Business Review and Forrester from October 2025 to September 2026. The median sample size was 410.

At the low end, The Wall Street Journal covered the NABE Outlook Survey, which polled 44 economists. At the large end, the Journal covered a Robert Half survey of 2,000 hiring managers. The largest studies all surveyed broad audiences such as hiring managers and small business owners, who are easier and cheaper to recruit. Studies of CEOs, CFOs and board members mostly ran between 100 and 600.

For a published report, a sample size of 400 is a good rule of thumb.

For a report you plan to publish, a good rule of thumb is 400 qualified respondents. With a sample of 400, the margin of error is 5 points, which is commonly accepted. Doubling to 800 would tighten the margin to about 3.5 points, and for many B2B reports that improvement isn't worth twice the cost. Knowing what the margin of error is for any given sample size can help you land on the amount that you and your stakeholders are comfortable with.

It's also important to pay attention to subgroups, or the type of data cuts you need. If you want your report to have data for, say, CTOs, but you also want to compare CTOs based in Austin vs. San Francisco, ideally you'd have at least 100 CTOs in Austin and 100 CTOs in San Francisco respond to your survey. With a sample of 100, the margin of error is 10 points, so “60% of Austin CTOs agree” should be read as 50% to 70%. Thirty is the lowest subgroup size to consider, and that would provide only directional guidance. With a sample of 30, the margin of error is 18 points, so 60% of Austin CTOs agree should be read as 42% to 78%. That range is too wide to quote as a precise figure, so treat a subgroup of 30 as a signal about which way the group leans. If a subgroup matters to your story, set a size quota for it before fielding so you know it will fill.

A brand tracker needs a larger sample to prove a change is real.

A brand tracker repeats the same survey every quarter or every year so you can see whether awareness or preference has changed. Every wave has its own margin of error, so when you compare two waves you have to allow for both. That's why a tracker needs a larger sample than a one-time survey to prove a change is real.

Here's an example. Your brand tracker runs with 200 respondents per wave. With a sample of 200, the margin of error is 7 points. Say 30% of your 200 respondents recognize your brand in the first quarter. You should read that as 23% to 37%. In the second quarter, 38% of a new group of 200 respondents recognize your brand. You should read that as 31% to 45%. Those two ranges overlap. Awareness could have been 35% in both quarters and produced both readings, so you can't tell whether awareness grew or you reached a different mix of people.

Now run the same tracker with 400 respondents per wave. With a sample of 400, the margin of error is 5 points. Say 30% recognize your brand in the first quarter. You should read that as 25% to 35%. In the second quarter, 40% recognize your brand. You should read that as 35% to 45%. No single figure fits both readings, so awareness went up and you can confidently report that.

Whichever sample size you choose, keep everything about the tracker the same from wave to wave. Use the same sample size, the same audience definition and the same screening questions in every wave. If the profile of the people answering changes, the results will change too, and you won't be able to tell whether your brand metrics shifted or your sample did.

Quick reference

Frequently asked questions

Why are 95% and 0.5 the standard settings?

A 95% confidence level is the convention in published research, and journalists and reviewers expect to see it. An estimated proportion of 0.5 produces the widest possible margin of error. Planning with 0.5 means the margin you plan for is the worst case, and your real margin can only be narrower.

Can I publish a study with fewer than 100 respondents?

Yes. Surveys of professionals are regularly published with 40 to 75 respondents. Just be clear about what the numbers can support. With a sample of 50, the margin of error is 14 points, so 60% agree should be read as 46% to 74%. Present the results as the view of that specific group, and don't break a sample that small into subgroups.

Does a bigger sample fix a poorly targeted one?

No. Margin of error only covers the randomness of who happened to answer. It says nothing about whether the right people answered. If a third of your respondents claim a title they don't hold, a sample of 2,000 is still wrong, and its tight margin of error will make it look more reliable than it is. Screening your respondents carefully does more for accuracy than adding more of them. That's one reason anti-fraud measures are such an important step of B2B recruiting.

Should I report the margin of error in my published study?

Yes. State the sample size, the confidence level and the margin of error in the methodology note, for example “400 US CTOs, 95% confidence, margin of error ±5 percentage points.” Journalists and analysts look for it.

Emporia Research sources verified B2B and healthcare professionals for quantitative studies. If you're preparing to run a study and need help with your sample plan, get in touch.