Why Going Is Different From Answering a Poll

Messaging Features vs Event Features August 22, 2026 13 min read By the Ontaym Editorial Team

Twelve people voted for that Saturday, so you booked a table for twelve, and seven turned up. Nobody lied to you. The poll answered a question about diaries, and you read the answer as a headcount.

The poll said twelve

You ran a poll. Three Saturdays, one winner, twelve votes for the middle one.

So you booked a table for twelve. On the night, seven people came.

Nobody lied to you. Nobody was rude, or flaky, or secretly hoping you would suffer.

The twelve was never a headcount. It was twelve people confirming that a particular Saturday did not currently clash with anything, which is a completely different fact wearing the same clothes.

This is the most common measurement error in group planning, and it is almost invisible because the two things look identical on screen. A poll option and an RSVP button are both a tappable row with a tally next to it. The interface gives you no reason to suspect they mean different things.

They mean very different things. One is about your calendar, the other is about your behaviour.

Availability, intention, commitment

There are three separate signals hiding inside "yes", and most groups only ever collect one of them.

Availability is a fact about your diary. On that Saturday, nothing else is written down. It is checkable, close to objective, and it says nothing about whether you want to go.

Intention is a fact about your current mood. You would like to be there, based on how you feel about the plan today, at this distance, with this much information.

Commitment is a promise. You are telling the organiser they can count you, spend money on you, and be annoyed if you do not show.

A poll collects availability. Sometimes it accidentally catches a little intention, because people do tend to vote for the dates they actually fancy.

It never collects commitment. Not because polls are badly built, but because commitment is not a question the poll asked.

Voting that Saturday works is a statement about your calendar. Saying you are going is a promise about your behaviour.

Look at the middle column and the problem states itself. The poll answered the first question beautifully, and you used it to do the third question's job.

The restaurant is where organisers find out

The realisation almost always happens in a restaurant, and it happens in a very specific way.

You arrive first, because you are the organiser and that is the deal. The host walks you to a table laid for twelve, which looks enormous and slightly hopeful.

Two friends arrive. Then two more, at a nice pace, and you feel fine.

At twenty past, you are at seven. The four empty places have that particular blankness of cutlery nobody is coming for.

Someone says the thing. "Wasn't Priya coming?" Priya voted for this Saturday six weeks ago and has not thought about it since, because voting for a Saturday is not the same as having plans on it.

Then the host appears with a friendly question about whether the rest are on their way, and you have to say a sentence you do not enjoy saying. You are not embarrassed because you counted wrong. You are embarrassed because you counted a thing that was never a count.

Why the gap widens as the date gets closer

Here is the cruel bit. A poll is at its most accurate on the day you run it, and it decays from that moment on.

When you vote, you are answering about a version of your life that has not happened yet. The Saturday is empty because nothing has been written on it, and things get written on Saturdays.

Six weeks is roughly forty chances for something to be added to that day. A wedding, a shift swap, a parent visiting, a train strike, a new relationship that eats every weekend.

None of those events know your poll exists. That is the actual failure: a poll result cannot update itself when reality moves, and reality moves constantly.

Compare that with a commitment. If someone said "yes, I am coming" and then a wedding invitation arrives, they feel the collision. There is a promise sitting there to be broken, and most people will message you, because a broken promise is a social event in itself.

A poll vote produces no such feeling. Nobody has ever apologised for a stale poll vote, because voting never felt like taking on an obligation.

That last row is the one that catches everybody. In a poll, silence and refusal are the same shape, which is why organisers spend so much time guessing.

Group size makes it worse, quickly

With four people, none of this matters. You know all four, you will speak to all four in the next week, and any drift gets caught in conversation.

At around a dozen, the maths turns on you. Each person carries their own small chance of drifting away from a vote they cast weeks ago, and those chances add up quietly across the whole list.

You cannot see the total, because everyone's reason is private and reasonable. From the inside it looks like one person dropping out, twice, which is normal. From the organiser's chair it is a table for twelve with seven people at it.

Bigger groups also have more people who never voted at all. In a group of forty, a poll with twelve votes tells you almost nothing about the other twenty-eight, and organisers read that silence far too generously.

We wrote about this scale effect from the platform side in the piece on what a Telegram poll can and cannot tell you about attendance, and it holds across every app that offers polls. The mechanics of the poll are not the variable. The variable is what a vote was ever able to mean.

The interface is doing this to you

It is worth being fair to organisers here, because this is not carelessness. Every tool you have hands you a number and lets you treat it as a headcount.

A poll shows a tally. A thumbs-up reaction shows a tally. A "who's in?" message with twelve replies feels like a tally, and none of these have any idea what they are counting.

The tally is the trap. A number implies a measurement, a measurement implies a thing measured, and your brain fills that in with "people who will be there" because that is the number you wanted.

Reactions are the purest example of this, and we gave them their own article: why emoji reactions are not attendance records covers exactly how a thumbs-up gets promoted into a guest list. A poll is the same failure with a more official-looking interface.

The more official the interface, the more you trust the number. That is precisely backwards, because the poll's authority comes from being good at a different question.

What a real commitment field looks like

If you want to see what it looks like when somebody designs a commitment field on purpose, the calendar standard is the place to look. It has been sitting there since 2009.

RFC 5545, the IETF specification behind every .ics file you have ever opened, defines an event as a set of named fields. One of those fields is attached to each attendee and is called PARTSTAT, short for participation status.

Its values are not a tally. They are the answer one named person has given, and the specification is careful about the difference.

Three things are worth stealing from that list even if you never touch a calendar file.

First, the answer belongs to a person, not to a pile. You are not looking at "nine", you are looking at nine names, each with a state you can ask about individually.

Second, there is a value for "I have not answered". NEEDS-ACTION is the default, which means silence has a name and shows up as an open item rather than blending into the background.

Third, "probably" is a legitimate answer with its own slot. TENTATIVE lets an honest maybe stay honest, instead of being rounded up into a yes by an organiser who needs a number for the restaurant.

None of that is available in a poll, and none of it is a flaw in polls. A poll was designed to choose between options, and it does that job well.

Polls are genuinely the right tool, for one job

This is not an argument against polls. It is an argument against asking a poll to be a guest list.

Picking a date out of three candidates is a real problem, and it is exactly the problem a poll solves. Ten people can answer in eight seconds each, nobody has to read a thread, and the winning option is obvious without anyone tallying by hand.

Try doing that conversationally and you will regret it immediately. Twenty messages later somebody says "whatever works for everyone" and the plan quietly dies.

So keep the poll. Use it for choosing between options, which includes dates, venues, cuisines, times, and whether the whole thing should be lunch instead.

Then close it and treat the result as what it is: a decision, not a headcount. The moment the poll picks a winner, its useful life is over.

The rule: stop polling when the plan becomes a thing

Here is a clean line you can actually use.

Poll while the event is still hypothetical. Ask for commitments the moment it becomes real.

An event is hypothetical while any of the big three are still open: the date, the place, or whether it is happening at all. During that phase nobody can commit, because there is nothing yet to commit to.

It becomes real the moment one of those gets fixed. Usually that is the date, and usually it is fixed by the poll itself, which is why the transition is so easy to miss.

The tell is money and space. If you are about to spend either on behalf of the group, book a table, buy tickets, order food, hire a room, you need commitments and a poll cannot give you them.

The other tell is the phrase "so we're all set for the sixteenth". If somebody in the thread has said that, the hypothetical phase ended and nobody announced it.

1. Poll the options, and only the options. Three dates, or three venues, never both in the same poll. Give it a deadline in the question itself so it does not drift.
2. Close it and say the answer out loud. "Sixteenth it is." A poll result that is never announced keeps collecting votes from people who think the decision is still open.
3. Ask a new, different question. Not "does that work?" but "are you coming?" The wording matters more than it should, because the first one is still asking about diaries.
4. Collect the answers per person, not as a tally. You want a list of names with states next to them, including the people who have said nothing.
5. Chase only the blanks. With per-person answers you know exactly who to nudge, which is usually three people rather than a broadcast to forty.
6. Re-ask when the plan changes. Any change to date, time or place invalidates every commitment, and pretending otherwise is how you end up with a table for twelve again.

Step six is the one people resist, because re-asking feels like admitting the plan is shaky. It is not: it is the only honest thing to do after you have changed the terms of the promise.

We have a whole piece on doing that gracefully, because people changing their RSVPs is normal rather than a failure, and a system that treats every change as a crisis will teach your friends to stop telling you.

Why "the poll said twelve" survives anyway

Organisers know all of this, at some level, and still book for twelve. It is worth understanding why, because the reason is not stupidity.

A poll number is available, and a commitment number requires work. Asking twelve people a direct question and tracking their answers is genuinely more effort than reading a tally, especially when the tally is sitting right there looking like an answer.

There is also a social cost to asking properly. "Are you actually coming?" carries a faint accusation, and in a friendly group that feels like a strange thing to say out loud.

That awkwardness is precisely why the poll gets misused. It lets the organiser get a number without asking anyone to promise anything, which is comfortable right up until the restaurant.

The fix is not more nerve. It is asking in a form where answering is small, changing your answer is small, and nobody has to be interrogated in front of the group.

The three sentences that quietly mean different things

Most of this trouble hides inside ordinary phrasing. Groups use these interchangeably and they are not interchangeable at all.

"That date works for me"

Availability. It reports the current state of a calendar and promises nothing. Perfectly honest, and useless for booking.

"I'm up for that"

Intention. It reports enthusiasm at a distance, which is real but weakly connected to behaviour. This is the one that gets over-read the most.

"I'm coming"

Commitment. It creates an obligation, and it is the only one of the three that a person feels the need to retract if things change.

Read your last planning thread with those three labels in hand. You will usually find a lot of the first two and almost none of the third.

That is not a group of unreliable people. That is a group that was never asked the third question, because nothing in the thread was shaped like asking it.

What a poll cannot do, no matter how good it is

Even a well built poll has structural limits that no feature can remove.

It closes. Once it has closed, it stops being a live picture of anything, and the number sits there looking equally confident on the day and six weeks later.

It has no concept of the event. A poll knows about options and votes, not about a Saturday, a restaurant and a booking, so it cannot notice when the plan moves under it.

It cannot represent absence. Someone who never voted and someone who voted no occupy the same amount of space in the result, which is none.

It flattens people into a number by design. That is a feature when you are choosing a date and a serious problem when you need to know whether to keep Priya's seat.

And it usually lives inside a conversation, which means the result scrolls away. Two weeks later the poll is a thousand messages up, and the number in everyone's head is whatever they remember it being.

Collect commitments somewhere that can change its mind

The constructive version of all this is simple. Keep polls for decisions, and give commitments somewhere of their own to live.

That somewhere needs three properties, and they are not exotic. It needs to hold one answer per named person, it needs to let that person change their answer without a conversation, and it needs to still be there on Thursday when you need the number.

Notice that a chat thread fails all three. A record of answers scattered through messages is not a list of names, changing your answer means posting again and hoping the organiser notices, and none of it survives contact with a busy week.

This is the same distinction we drew in the difference between an RSVP and a reaction in a group chat, applied to polls instead of emoji. The pattern repeats because the underlying mistake repeats: a lightweight social signal gets promoted into a commitment record it was never built to be.

You can do this by hand if you like. Plenty of organisers keep a note on their phone with names and states, updating it as people reply, which works well and makes one person into the database.

That is a real cost, and it falls on the same person every time. If planning in your group has quietly become one person's admin job, the piece on why a group chat is not the same thing as an event explains what that person is actually doing manually.

What to say instead

Wording does a surprising amount of the work here. The question you ask shapes the answer you get, and most planning questions are accidentally asking about calendars.

The right-hand column has a tone to it, and some people find it too blunt for a friend group. In practice it reads as organised rather than bossy, because it tells everyone that a real decision has been made.

The magic phrase is the one that names what the number is for. "For the table" or "for the deposit" turns an abstract question into an obvious one, and people answer honestly when they can see the consequence.

Where Ontaym fits

Ontaym exists for the second half of this, which is the honest reason this article is on our blog.

An Ontaym event is a record with one answer per person on it, and those answers can change up to the moment the thing starts. Guests answer at a link without installing anything or handing their phone number to everyone else on the list.

Your poll stays exactly where it is. Run it in whatever chat your group already lives in, decide the date, and then collect the commitments somewhere that can keep up with reality.

That is a deliberately narrow job. It will not help you choose between three restaurants, and it is complete overkill for coffee with one friend.

The one thing to take away

Next time a poll closes, do not write the winning number down as a headcount. Write down the date, and then start again.

The second question is short and it is the only one that matters: now that it is definitely the sixteenth, are you coming?

You will get a smaller number. That smaller number is the true one, and it is the one the restaurant should hear.

Everything else follows from taking that seriously. Ask per person, let people change their minds cheaply, chase the blanks rather than the group, and re-ask whenever the plan moves.

Do that and the table stops having empty places in it. Not because more people came, but because you finally counted the right thing.

Frequently asked questions