Yes/No counter · Free · No sign-up
Yes / No Counter
A yes/no counter tallies two options at once and shows the split between them. Each key records one response, and the percentage updates as you go. This one shows the sample size beside the percentage and withholds a confidence range until there is enough data for one to mean anything.
Free, no sign-up, works offline, and no respondent data is collected.
Y and N count the two options, ← and → do the same, Ctrl+Z undoes.
This counts and divides. It does not make a bad sample good. The interval it shows assumes you drew people at random from the group you want to talk about; tallying whoever walks past is not that, and no arithmetic here can repair it. Nothing is sent anywhere and no respondent data is collected, because only two running numbers exist.
Any two-way counter gives you 31 and 19. The useful question is what the 62% is worth, and that depends on two things almost no tool shows you: how many you counted, and whether the people you counted were a random sample at all.
A yes/no counter is a two-category tally counter used for surveys, polls and quick head counts, recording how many chose each of two options and the proportion between them.
- No sign-up
- Free to use
- Counts never leave your browser
- Works offline
- Unlimited counters
Key facts
| Detail | Answer |
|---|---|
| Counts | Two options at once, each with its own key |
| Labels | Editable. Yes/No, Agree/Disagree, Pass/Fail, In/Out |
| Shows | Count per option, total, and the live percentage |
| Sample size | Displayed beside the percentage, not beneath it |
| Confidence range | 95%, from n = 30 on |
| Random sample | Asked for explicitly, not assumed |
| Undo | Yes, last 100 changes |
| Offline | Yes, after the first load |
| Respondent data | None. Two running numbers and two labels |
How to run a two-option tally
The counting takes a second. Deciding what the number is worth is the part worth doing properly.
Name the two options
Yes and No by default, but Agree and Disagree, Pass and Fail or In and Out work the same. Naming them stops you losing track of which key was which.
Decide how you are choosing people, before you start
Whether you will ask everyone who passes, every fifth person, or a list drawn at random. This decides what your final number can be used for, and it cannot be fixed afterwards.
Tick the random box only if that is true
It is a claim about your method, not a setting that improves the data. Unticked, the page still shows the range but keeps the condition attached to it.
Count with the keys
Y and N, or the arrow keys, or the two buttons. Undo takes back a mis-tap without disturbing the other option.
Keep going past thirty if you can
No range is shown below thirty, because one calculated there is more confident than the count deserves. The range narrows quickly after that.
Copy the result with its method
The copied summary carries the counts, the share, whether the sample was random and the interval, so the number never travels without its caveats.
Where a yes/no counter gets used
Anywhere a question has two answers and you need the split rather than a total.
Quick opinion surveys
Asking a yes/no question at a stand or a door and wanting the share, not just the count.
Pass and fail checks
Inspections, quality checks or practice attempts, where the proportion passing is the figure that matters.
A/B preferences
Two samples, two options, one key each. The labels take whatever the options are called.
Classroom votes
A show of hands counted honestly, with the total visible so nobody has to trust a rough estimate.
Attendance and no-shows
Turned up against did not, where the percentage is the thing being reported upward.
Sports and practice attempts
Made against missed, giving a live success rate over a session rather than at the end of it.
Why the percentage is the least useful number on this page
Every two-way counter will give you 31 and 19. Divide and you get 62%, and the percentage is the number that ends up in the email. It is also, on its own, close to meaningless — and the two reasons why are the reason this page exists rather than being a preset on the multi counter.
A percentage without n is not a result
Two out of three is 67%. Two hundred out of three hundred is also 67%. They are not the same claim, they are not close to the same claim, and written as a bare percentage they are indistinguishable. This is why the sample size sits next to the percentage here at nearly the same size, rather than in grey text underneath it. How big n must be for a given margin is worked out in the guide to how many people to survey.
| Counted | Share | 95% range, random sample | What you can say |
|---|---|---|---|
| 5 of 8 | 62.5% | not shown | Nothing. One more tap moves it 7 points |
| 19 of 31 | 61.3% | ±17 points | Possibly a majority, possibly a minority |
| 62 of 100 | 62% | ±10 points | Probably a majority |
| 620 of 1000 | 62% | ±3 points | A majority, and roughly this big |
Read the second row again. Thirty-one people, a clear-looking 61%, and the honest range still reaches from 44% to 78% — it does not even settle whether most people said yes. That is the count most informal surveys actually stop at, and it is why no interval at all is shown below thirty here: a tidy ± on twelve responses reads as precision and is not.
The interval only means anything if the sample was random
This is the bigger of the two problems and the one almost nobody states. The arithmetic behind that ± assumes every member of the group you want to describe had an equal chance of being counted. Tallying whoever walks past a table does not do that. You sampled people who were in that place, at that time, and willing to stop — which is a real group, but not the group you meant.
That is a convenience sample, and the honest thing to do with one is report the count and how you collected it, not a confidence interval. So the checkbox is unticked by default, and while it is unticked the page shows the range with the condition attached rather than quietly implying it holds. Ticking it is a claim you are making, not a setting that improves the data.
Three ways a count that feels fair usually is not: people who feel strongly answer and the indifferent walk on, the question is asked differently once you are tired of asking it, and whoever is in that corridor at 3pm is not whoever is in it at 9am. None of these show up in the arithmetic. All of them move the answer more than the ± does.
What this does not do
It holds two categories, not five, and it deliberately stops there: a third option turns a proportion into a distribution and the honest statistics change with it. Use the multi counter for more categories, then the tally to percentage calculator to turn those counts into shares that add to 100, and the chart maker to draw them, with the same caution about the range. It does not record who answered, does not weight or adjust for who you did not reach, does not do a chi-squared or any test of significance, and does not know anything about your population size. Everything stays in this browser; the only things stored are two numbers and your two labels.
What a margin of error covers, and what it leaves out
The range shown beside the percentage is a 95% margin of error. Pew Research Center explains what that promises in its guide to the margin of error in election polls: a margin of plus or minus 3 points at the 95% level means the result would be expected to fall within 3 points of the true value in 95 of 100 samples drawn the same way. The example it gives is a simple random sample of 1,067 people.
Pew is equally plain about the limit: the margin reflects sampling variability only, and is “only one of many possible sources of error”. It names gaps in who can be reached, respondents not answering candidly, and differences in question wording as errors that are much harder to quantify and rarely reported. None of them is in the range this counter shows, which is why the random-sample box asks about your method instead of assuming it.
Below 30 responses the counter shows no range at all. The formula it uses is the normal approximation, p ± 1.96 × √(p(1 − p) ÷ n), and the Wikipedia article on the binomial proportion confidence interval notes that this approximation becomes unreliable when the sample size is small, and when the true proportion is close to 0 or 1. So when fewer than five people gave either answer, the counter shows the Wilson score interval instead, as a range rather than a ±, because near 0% or 100% the ± formula shrinks towards zero. A tally of twelve people can be counted exactly; what it cannot do is bound the opinion of everyone else.
Online counter vs a clicker, an app, paper and a spreadsheet
Every counting method trades something away. This is what each one actually costs you.
| Feature | This online counter | Handheld clicker | Counter phone app | Pen and paper | Spreadsheet |
|---|---|---|---|---|---|
| Cost | Free | $5–$15 | Free with ads, or paid | Cost of paper | Subscription |
| Anything to install | No | Buy the device | App store download | No | Yes, or a web login |
| Several counts at once | Unlimited | One per device | Usually limited | Yes, messy | Yes |
| Fix a miscount | Undo and redo | Start over | Sometimes | Cross it out | Edit the cell |
| Counts by more than one | Any step size | One at a time | Sometimes | Manual maths | Yes |
| Below zero | Yes | No | Rarely | Yes | Yes |
| Keeps the count if you close it | Saves automatically | Until you reset the dial | Yes | If you keep the paper | Yes |
| Works with no signal | Yes, after first load | Yes | Yes | Yes | Often not |
| Keyboard and shortcuts | Full | None | None | None | Partial |
| Sends your counts anywhere | Never | Never | Often, for accounts and ads | Never | Yes, to the cloud |
Other counter types
The same counter, set up and explained for a specific job. Your counters are shared across all of them.
Everyday6
The general-purpose counters: one number, many numbers, or counting on a timer.
Sticker album counter
Stickers you have and spares to swap, missing lists to copy, and the packs still to go from your count.
Open sticker album counter →Online counter
The all-purpose counter. Any step size, any number of counters, negative values allowed.
Open online counter →Tally counter
Tally marks in groups of five for surveys, stock takes and wildlife counts.
Open tally counter →Click counter
A browser clicker that replaces the handheld thumb counter for door headcounts.
Open click counter →Multi counter
Run as many counters as you like at once, side by side in a grid.
Open multi counter →Auto counter
Counts on its own from every 0.1 seconds to once an hour, and stops at your limit.
Open auto counter →
All 72 counters, by category: Everyday (7) · Sport & scores (44) · Fitness & health (7) · Work & field (5) · Faith (9)
Frequently asked questions
What is a yes/no counter?
A tally counter with two keys instead of one, so you can record two answers to the same question and see the split between them. It is the standard tool for a quick survey, a pass/fail check or any count where the proportion matters more than the total.
How do I count two things at once?
Give each option a key and tap the one that matches each response. Here that is Y and N, the arrow keys, or the two buttons. Each option keeps its own running count, and the percentage between them updates on every tap.
How many people do I need to survey?
More than feels necessary. At 31 responses a 61% result still spans roughly 44% to 78%, which does not even establish a majority. A hundred gets you to about ten points either way, and a thousand to about three. No range is shown here below thirty.
What is a margin of error?
How far the true proportion could reasonably sit from the one you measured, given how many you counted. It shrinks as the sample grows, and it only describes sampling luck. It says nothing about a leading question or about the people you never asked.
Why does the counter ask whether my sample is random?
Because the confidence range is only meaningful if it was. The arithmetic assumes everyone in the group you want to describe had an equal chance of being counted, which is not true of whoever happened to walk past. Unticked, the range is shown with that condition attached.
Is counting people as they walk past a random sample?
No. That is a convenience sample: you have counted people who were in that place, at that time, and willing to stop. It can still be worth doing, but report the count and your method rather than a margin of error.
Can I change the Yes and No labels?
Yes, to anything up to 24 characters. Agree and Disagree, Pass and Fail, In and Out, or two product names all work, and the labels appear on the keys and in the copied result.
Can I count three or more options?
Not here, deliberately. A third option turns a proportion into a distribution and the statistics shown on this page would no longer be the right ones. The multi counter runs as many independent counters as you need.
Does it save my count if I close the tab?
Yes. Both counts and your labels are written to this browser's local storage as you tap, so closing the tab or locking the phone loses nothing. Clearing site data clears them.
Is this a yes or no generator?
No. A yes or no generator picks an answer at random, often with a tally of what it has picked. This page counts the answers people actually give: one key for each option, the split as a percentage, and how many answered. To let chance decide instead, toss a coin.
Can other people vote on it through a link?
No. The two counts live only on the device you tap them on, which is why no respondent data exists anywhere. Use it to count answers as they are given, such as a show of hands in a room; the guide to counting votes at a meeting covers the methods and the checks. For votes cast from other people’s phones, use a poll tool that collects responses.
Is any respondent data collected?
None at all. Only two running numbers and two labels exist, all held in your own browser. There is no record of individual responses because none is ever created, and nothing is sent to a server.
Counting guides
Longer answers to the questions behind the counting.
- Surveys
How many people do you need to survey?
Why 385 answers a city and 45 answers a club, and the margin each sample size buys.
- Counting
How to count votes at a meeting
Majority means more than half of the votes cast, not 50% plus one, and an abstention is not a vote. The four voting methods, two-teller counting, and a tellers' report worked with numbers.
- Ecology
Mark and recapture: Lincoln-Petersen, Chapman and the confidence interval
Both estimators worked through, which confidence interval fits your counts, the five assumptions, and what a handful of recaptures does to the answer.
- Surveys
How to tally survey responses by hand
A worked tally of 60 paper forms, from the sheet to the percentages. Blank answers, two boxes ticked, tick-all questions, write-ins, and the check that proves the count.
- Tallying
Tally marks explained
Groups of five, the five-bar gate, and why the convention has survived for centuries.
- Reference
Types of counters explained
Synchronous vs asynchronous, mod-N, ring and Johnson, industrial and PLC counters, plus mechanical and software counters.