Survey
Course feedback survey builder
Short, research-backed questions for the middle or the end.
Free tool
Training feedback calculator
How sure can you be when 12 of 18 learners say a lesson was clear? Get the honest range for any yes/no question, NPS with its margin of error, the average of 1–5 ratings, and how many answers you need.
Free, no sign-up. Nothing you type leaves your browser.
Your numbers
Agreed, found it clear, would recommend…
Everyone who answered the question
Result
67% likely 44%–84%
Most of the group, give or take 20 points.
A rough direction, not a verdict: the likely range is wide.
How to use the calculator
- Yes / agree: enter how many said yes (agreed, found it clear, would recommend) and how many answered. You get the share, the likely range and, for small groups, the same result in words.
- NPS: enter promoters (9–10), passives (7–8) and detractors (0–6). You get the score and the margin of error that NPS reports usually leave out.
- Ratings 1–5: enter how many people chose each rating. You get the average with its range and the share of 4s and 5s.
- Answers needed: choose the margin you can live with and, optionally, the size of the group. You get the number of answers that margin takes.
Why small groups need a range, not a percentage
Training groups are small: 8 to 30 people is typical, and not everyone answers. With 18 answers, one person changing their mind moves the result by more than 5 points. So “67% found the lesson clear” really means “somewhere between about 44% and 84%”. Report it that way and nobody rewrites a lesson, or celebrates, on the strength of two answers.
The range comes from the Wilson score interval at 95% confidence. It behaves well where the textbook formula breaks down: with few answers, and with results near 0% or 100%, where the textbook formula can give a range below 0% or a range of zero width.
A worked example
You ran the same workshop for two groups. In group A, 12 of 18 said they could explain the topic to someone else (67%, likely 44–84%). In group B, 9 of 20 said so (45%, likely 26–66%). The ranges overlap from 44% to 66%, so the difference may be chance. That doesn’t mean “do nothing”: look at what the lesson asks and where people said they got stuck, and check again with the next group after you’ve changed it.
What about NPS?
Net Promoter Score subtracts the share of detractors from the share of promoters, so it carries the uncertainty of both. With 20 answers its margin of error is around ±34 points. Most course NPS figures are reported without that margin, which is how a group “drops 20 points” without anything having changed. If a client asks for NPS, give it with its range. If you’re choosing the question yourself, a yes/no question after each lesson or a labelled 1–5 rating at the end gets more answers and says more.
Ask better questions, get readable answers
- Ask everyone at the same moment. Voluntary feedback skews to the very happy and the very unhappy. A one-tap question at the end of every lesson reaches the people in between.
- Ask about performance, not feelings. “Could you explain this lesson to someone else?” predicts understanding better than “Was this clear?”, which mostly measures how fluent the text felt.
- Keep it short. Every extra question lowers how many people finish. If you need a template, the course feedback survey builder starts from short, research-backed questions.
Questions people ask
How many answers do I need for course feedback to mean something?
Below 5 answers, nothing: report the count only. From 5 to 10, say it in words (“about half”, “most”) rather than as a percentage. From around 30 answers a share is usually within ±15 points, which is enough to compare one group with the next. A margin of ±5 points needs about 385 answers, more than most courses ever get.
What is the margin of error of a survey?
It is how far the result could be from the true value for everyone like the people who answered, 95% of the time. This calculator uses the Wilson score interval, which stays honest with small groups and results near 0% or 100%, unlike the textbook “±1.96 × standard error” formula.
Is NPS a good measure for training courses?
It’s popular, but it’s noisy for small groups: with 20 answers the margin of error is about ±34 points, so a score of +30 could just as well be −4 or +64. For a course, the share who say “yes, I could explain this to someone else” after each lesson, or a 5-point rating with a range, tells you more for the same effort.
Why does the result say “about half” instead of a percentage?
With fewer than 10 answers a single person moves the result by more than 10 points, so a percentage looks more precise than it is. Words keep everyone, including you, from reading too much into it.
How do I compare two groups?
Calculate the range for each group. If the ranges overlap a lot, the difference may be chance. If they barely overlap or don’t, the difference is probably real. With small groups, compare the same question asked at the same moment, for example at the end of the same lesson.
Is my data stored anywhere?
No. The calculator runs in your browser; nothing you type is sent to a server or saved.
Do this for every lesson, every group
Chalkenso is where courses are designed and improved: instructors note what happened on the exact passage, learners say where they got stuck, and AI drafts the next version for you to approve.