Discord Events: What Happens After Someone Says Interested?

Discord Communities & Events August 22, 2026 14 min read By the Ontaym Editorial Team

Forty-two people pressed Interested and nine turned up. Nobody lied and nothing is broken, because Discord's own documentation describes that button as a subscription rather than an answer. The whole problem lives in the gap between what a subscriber count measures and what an organiser reads into it.

Forty-two Interested, nine chairs

A board game night, a Tuesday, a bar upstairs room that holds thirty. The server has a Scheduled Event with a nice cover image and forty-two people marked Interested.

Nine turn up. The organiser has paid for the room, bought four bags of crisps nobody eats, and spends the evening quietly working out what she did wrong.

The answer is that she did nothing wrong. She read a number correctly and then asked it a question it was never built to answer.

This article is about one button. Not Discord as a whole, not community building, just the small blue control that says Interested and the extraordinary amount of confusion it generates.

The interesting thing is that the button is honest. It says exactly what it does, in plain English, and it is the reader who supplies the wrong meaning.

What the button is called on the inside

Start with the part almost nobody looks at. Under the interface, Discord's own developer documentation names the thing precisely.

A Scheduled Event carries a field called user_count. Discord's documentation for the guild scheduled event object defines it as the number of users subscribed to the scheduled event.

Subscribed. Not attending, not confirmed, not accepted. Subscribed.

That word is not an accident or a bit of sloppy engineering shorthand. It runs consistently through the whole API surface.

The realtime events are named the same way. Discord's gateway event reference describes Guild Scheduled Event User Add as sent when a user has subscribed to a guild scheduled event, and User Remove as sent when a user has unsubscribed.

Subscribed and unsubscribed. Those are the verbs of a mailing list, not a guest list.

The API is telling you it built a notification preference. The organiser keeps hearing a headcount.

The interface wording is gentler and it points in exactly the same direction. Pressing the button tells the host the event appeals to you, and it signs you up for a notification when the event goes live.

Those are two very different things sitting in one sentence. One is a soft social signal, and one is a concrete mechanical outcome, and only the mechanical half is what actually gets stored.

You can also un-press it, quietly, at any point. Discord emits an unsubscribe event when you do, and a thing you can silently retract was never being sold as a commitment.

There is an endpoint that lists the subscribed users, so an organiser can at least see who they are. That helps with names. It does not turn any of those names into a yes.

In defence of everybody who pressed it

Here is where most articles about this get unpleasant, and where this one is going to stop and be fair.

Pressing Interested is not flaky behaviour. It is a small, generous, completely honest act.

Think about what a person is doing in that half second. Something in a server they like has appealed to them, and they want to register that fact and be reminded later.

That is a real message and it is worth receiving. Someone who presses Interested has told you the idea works, the framing works, and the timing is at least plausible for them.

Nobody presses Interested cynically. There is no advantage in it, no social credit, and no audience watching to be impressed.

They are also being kind. In a channel where an organiser has just posted something they clearly care about, pressing Interested is the cheapest available way to say this is a good idea, do it.

The word "interested" is doing honest work in ordinary English too. If a colleague says they are interested in a job, nobody files that as an acceptance.

So the deception, if there is any, is not coming from the people pressing the button. It is coming from a number in an interface that looks exactly like a total.

Forty-two is a very convincing number. It sits next to a date and a venue, it goes up over time, and everything about its presentation suggests it is counting people who will be in a room.

What Interested reliably tells you

The useful move is not to distrust the signal. It is to work out what the signal is actually good for, and then use it for that.

Interested is a genuinely strong measurement of some things. It is simply measuring at the top of the funnel rather than the bottom.

The bottom four rows are the entire problem. They are also, unfortunately, the only four rows a venue cares about.

Notice how clean the top half is. Used as an appeal meter, Interested is one of the better instruments a community organiser has.

A server that ran six events last year and has the Interested counts for each is holding real programme intelligence. Board games got forty-two and the film night got seven, and that comparison is trustworthy even though neither number is a headcount.

Why online events hide the gap and offline events expose it

Now the part that explains why this feature has survived for years without anybody complaining much.

For an online event, the gap between Interested and attending barely matters. The two numbers can differ wildly and nothing bad happens.

Consider a voice chat on Sunday evening. If forty are Interested and twelve show up, the event is a success and cost nothing.

Nobody booked a room. Nobody paid a deposit. Nobody bought food that will be thrown away.

Drifting in halfway through is normal, and leaving without saying goodbye is normal. The event does not have a door and it does not have a capacity.

Then somebody points the same button at a bar with eleven stools. Every one of those forgiving properties inverts at once.

Read the right column and the picture is clear. The feature is well calibrated for the event type Discord actually runs most of, and every property that made it forgiving disappears when a physical room is involved.

The location filter deserves its own sentence. A server of four thousand is scattered across continents, and an Interested from two streets away is stored identically to an Interested from a different hemisphere.

Discord is not hiding this from you. It simply never collected the information, because for a voice channel there was nothing to collect.

The mistake is on the organiser's side, and it is a reading error

This is worth saying flatly, because it changes what you do next.

Nothing about Interested is broken. The failure happens in the half second where a human converts a subscriber count into an expectation.

That conversion feels automatic because of how the number is presented. It is big, it is near a date, and it goes up.

Compare it to a view count on a video. Nobody looks at eight thousand views and books a venue for eight thousand people, because the vocabulary keeps them honest.

Interested does not offer that protection. The word sits just close enough to attendance that a tired organiser at 11pm will round it up.

There is a second, subtler version of the error. The organiser does not think forty-two are coming, but they do think forty-two is the pool the attendees will come from.

That is also wrong, and in both directions. Some people who never pressed anything will absolutely turn up, and the pillar article on why Discord works for communities and struggles with offline plans covers why silence in a server carries almost no information.

The five-way tie that Interested creates

Here is the practical damage. Behind that single count sit at least five different people, flattened into one row.

• The person who is definitely coming and treats the button as their RSVP
• The person who wants the start notification and will decide on the day
• The person supporting the organiser socially and has no plan to travel
• The person on the wrong continent who just likes the sound of it
• The person who pressed it in March and has forgotten the server exists

All five produced the same increment. No column anywhere separates them.

And behind the people who pressed nothing sit at least three more: not seen, seen and declined, and seen and definitely coming but not a button-presser. If you want that dissected properly, the piece on the difference between server membership and real participation handles the wider version of the same problem.

What an answer actually requires

Interest and an answer are different objects, and the difference is not effort. It is grammar.

Interest

A statement about the event. It says this appeals to me, and it stays true no matter what you do with it.

Availability

A statement about your calendar. Tuesday is free, which is a fact about you rather than a plan involving anyone else.

Intention

A statement about the future. I mean to come, held privately, revisable without telling anybody.

An answer

A statement made to a named person about one specific date, which that person is entitled to plan around.

Only the last one can be booked against. It is the only one with another human on the receiving end.

That is what makes an RSVP feel heavier than a button press, and the weight is the point rather than a design flaw. A promise costs something, which is exactly why it predicts anything.

The calendar world formalised this decades ago. RFC 5545, published by the IETF in 2009, gives every attendee a PARTSTAT value: ACCEPTED, DECLINED, TENTATIVE, or by default NEEDS-ACTION.

Four states, one per person, with silence given its own name. Discord's model has one state, aggregated into a count, with no way to say no.

Its companion, RFC 5546 on scheduling requests and replies, treats the invitation and the reply as two halves of one exchange. Interested is a very good half of that exchange with the other half missing.

The missing decline is worse than it sounds

There is no Not Interested button, and that absence does more damage than the soft yes does.

A no is genuinely useful to an organiser. Eleven yeses and thirty nos is a clear picture, and eleven yeses out of a fog is not.

Without a decline, everyone who is not coming stays silent, and silence is unreadable. That is the same failure the article on why emoji reactions make terrible attendance records describes in a chat context, arriving here through a much nicer interface.

It also means a change of mind leaves no trace. Somebody who genuinely intended to come in March and cannot come in April has no way to update their entry, and un-pressing feels like a small public retreat.

Turning interest into an answer you can book against

None of this means abandoning the button. It means adding one step that Discord does not provide.

The pattern is simple: let Interested do the job it is good at, and ask the attendance question separately, in words, of named people.

1. Keep the Scheduled Event and stop treating the count as a headcount. It is the best announcement mechanism in the product, with a real date shown in each member's local time. Look at the number as an appeal score and nothing else.
2. Ask a closed question with a date in it. Not "who fancies board games", which invites interest. "Are you coming on Tuesday the 14th, yes or no" is the only phrasing that produces an answer.
3. Ask the people who pressed Interested, by name. You can list them, so use that list. A direct, warm, individual ask converts far better than another broadcast into a channel of four thousand.
4. Make no as easy as yes. Say out loud that a no is genuinely helpful and takes nothing away. Without that, the people who cannot come will go quiet, and quiet is the answer you cannot plan around.
5. Ask where they are before you ask if they are coming. One question about rough location, answered privately, removes the entire time zone illusion in a single move. Nobody needs a street, only a city.
6. Confirm again two days out. The drop between the first yes and the second yes is real and it is the number the venue should hear. Booking against the first yes is how you end up with four bags of crisps.
7. Keep the answers somewhere that survives the channel. A list of yeses that lives in message history will be unfindable by Friday, which is the point made in tracking real attendance for a Discord event.

Step five is the one experienced organisers add and beginners skip. It is also the one that quietly explains most of the gap in a large international server.

Step six feels like nagging and it is not. Asking twice is what turns a guess into a commitment, and everyone who says yes twice knows they have said something real.

What not to do instead

The instinct after a bad Tuesday is to reach for another lightweight tool. Most of them recreate the same problem in a new shape.

A poll is the usual candidate. Polls close a question at a moment in time, and attendance keeps moving afterwards, which the article on what polls measure versus what commitments measure works through properly.

Reaction roles are the other one. They are a beautiful mechanism and they still produce a count of people who pressed something, which is precisely the object you already had.

Adding a second soft signal does not average out into a hard one. Two hunches are still a hunch.

A note on scale, because it cuts both ways

Small servers sometimes read all this and conclude it does not apply to them. Usually they are right, and it is worth saying why.

In a server of thirty people who all know each other, Interested is nearly accurate. Everyone knows each other's names, most live near enough, and not showing up has a social cost.

The signal has not improved. The surrounding context has done the work the button cannot do.

Scale removes that context piece by piece. At four hundred members the names go, at four thousand the geography goes, and the social cost of a no-show goes somewhere around the point where you stop recognising handles.

So the rule of thumb is not about server size directly. It is about whether you could name, unprompted, the people you expect to see.

If you can, trust the count. If you cannot, the count is a mood reading and you need an actual conversation, which is roughly the argument in why "going" is a different thing from a poll answer.

Where Ontaym fits

This is the one section about our own product, and then it stops.

Ontaym holds the half of the exchange that Interested does not. An event is a record at one address, and each guest gives a real yes or no attached to their name.

You post that link in the same channel as the Scheduled Event, so the announcement keeps its reach and the answers stay private. Nobody joins anything, nobody makes an account, and nobody's location goes into a public channel.

When the time moves, you change it once and everyone who said yes is looking at the change rather than scrolling for it. That is the difference between a subscriber list and a guest list, expressed as software.

It is a deliberately narrow tool. A Sunday voice chat wants a Scheduled Event and nothing more, and a ticketed convention wants a ticketing platform.

The one sentence version

Interested is a subscription. Discord's own documentation says so in the plainest possible language, and the people pressing it are behaving exactly as invited.

An organiser who reads it as a headcount is doing something no button asked them to do. That is a fixable habit rather than a product complaint.

The fix is one extra question, asked with a date in it, of people you can name. It takes an evening and it converts a mood into a number.

Do that and forty-two Interested becomes something genuinely useful: a list of forty-two people worth asking a real question. Some of them will say yes, more than you expect will say no, and for the first time you will know which is which.

Then you can book the right room. The crisps are still your own problem.

Frequently asked questions