Replies are not RSVPs, and RSVPs are not attendance
One dinner, three promotion channels, three different funnels. Why the channel with the most replies produced no attendance, and what to measure instead.
- Reply volume and attendance are two different outcomes. The channel that produced the most replies for one dinner produced nobody at the door.
- A channel that a platform can interrupt is not a channel you can plan a room around, however well it performs while it is running.
- The largest single group walking into a side event is often people nobody can attribute to any channel, which leaves the result impossible to repeat.
A team promoting a dinner at a large fintech conference ran three channels in parallel. Social outreach, a marketing email to an existing list, and a partner co-branding the invitation with them.
Social outreach won on replies by a wide margin. On the night, none of the people who replied on social came.
The email list, which had produced the fewest replies of the three and the slowest, produced almost everyone who walked in. The partner list produced the rest. And the largest single group at the door was people nobody could connect to any channel at all.
The reflex after a dinner is to look at the channel that generated the most conversation and do more of it next time. Here that would have been the wrong call.
Three channels, three completely different funnels
There are four stages between an invitation and a person in the room, and each one is a real filter.
| Stage | What it measures | What it costs the recipient |
|---|---|---|
| Invited | Reach | Nothing |
| Replied | Interest | Thirty seconds |
| RSVP | Intent | A calendar decision |
| Walked in | Attendance | An evening at a conference, against every other invitation they hold |
Different channels perform differently at each stage, and a channel can be excellent at one and useless at the next. Social outreach is conversational, so it converts well from invited to replied. That is what it is good at. Converting a warm social reply into a calendar commitment is a separate job that the channel does not do for you.
Email is the reverse. It generates fewer replies and the replies come slower, and the ones that arrive tend to arrive as a decision rather than a conversation, because replying to an email costs more effort than tapping a message. A confirmation with a time, an address and a calendar file attached closes the loop in one step.
A channel that can be interrupted is not a channel you can plan around
The social campaign in that example did not underperform. It was stopped. Partway through the sequence the platform’s automation checks flagged the activity and the outreach was paused, which meant a set of warm replies never received the follow-up that would have turned them into confirmed guests.
Channel performance is not only about how well a channel converts. It is also about whether you control the channel.
Email running on a domain you own and authenticate can be planned around. It can be paced, scheduled, and finished on the date you chose. Outreach running inside somebody else’s platform, under rate limits and automation detection that neither publishes its thresholds nor warns you before acting, cannot be planned around in the same way. It may work well, and it may stop partway through without warning.
The practical conclusion is not to abandon social. It is to stop putting a fixed-capacity commitment on top of it. A dinner has a booked room, a deposit and a headcount. That commitment should rest on the channel that will still be running next week.
The largest group at the door is usually untracked
The unattributed group is the finding with the most planning consequence.
If the biggest single group of attendees cannot be traced to any channel, the result is not repeatable. Next time you will make the same three bets in roughly the same proportions, because there is no evidence pointing anywhere else.
The people in that bucket are real and mostly explicable: someone forwarded the invitation, a rep mentioned it at the booth, a partner brought a colleague, somebody saw it on a screen in the sponsor hall. All of that is fine, and none of it was recorded.
Attributing a side event is much easier than attributing pipeline, because the funnel is short and the guest list is finite. It needs three records, and a typical program keeps only one of them:
- A registration link per channel. Not one link everywhere. One per channel, so the RSVP itself carries its origin.
- The channel written onto the RSVP. Including for a forward. Ask how they heard about it on the form. One field.
- A door list. Somebody with the guest list checking people in. This is the record nobody keeps and it is the only one that measures the outcome you paid for.
Without the door list, every conclusion you draw is about interest rather than attendance.
What to measure instead of replies
Replies are worth reading and worth answering quickly. They are not worth reporting as the result of the campaign.
Three numbers describe a side event:
| Number | How to read it |
|---|---|
| RSVP rate per channel | Invitations to confirmed calendar commitments. The planning number |
| Show rate per channel | Confirmed invitees who turned up. Usually well under 100 percent, and it varies by channel more than people expect |
| Meetings created | Conversations at the dinner that produced a scheduled next step |
The third one is the only one that connects to pipeline, and it is the reason to run the dinner at all. If nobody leaves with a scheduled next step, the spend stays on the budget as hospitality. Capturing owner, problem and next step while the conversation is still specific is what moves it, which is the subject of running event dinners without vendor happy-hour energy and, on the floor side, in-event capture.
Planning capacity from a funnel you can see
Once you have the three numbers, the capacity question becomes arithmetic.
Work backwards from the room. Decide how many people you want at the table, divide by your show rate to get the RSVPs you need, divide by your RSVP rate to get the invitations you need to send, and check that number against the size of your enriched, ICP-matched attendee list for that event. If it does not fit, the constraint is the list rather than the invitation, and the fix is upstream in sourcing.
Two adjustments hold up in practice.
Send the invitation while the recipient’s evenings are still free. Senior attendees at a large conference commit their nights early, and an invitation arriving in the final week is competing for a slot that has already gone. Late invitations depress attendance more reliably than weak copy does.
Filter the invitation list as hard as you filter a sequence list. A dinner invitation that goes out broadly will get accepted by people who are delighted to attend and are never going to buy anything. They cost you seats. The same ICP scoring that decides who your reps chase at the booth should decide who gets the invitation, and the event selection scorecard covers the upstream version of the same judgement.
The invite sequence that survives all three problems
Putting it together, the version that holds up looks like this.
Start several weeks out, from an ICP-filtered attendee list rather than from everyone you can reach. Run two channels at minimum, with the fixed commitment resting on the one you control. Give each channel its own link. Confirm every yes within the day, with a time, an address and a calendar file, because an unconfirmed yes is not an RSVP. Send one reminder the morning of. Keep a door list.
Then, afterwards, write down the three rates. Besides the conversations themselves, those rates are what the dinner leaves you with, and they are what makes the next one a plan rather than a repeat of the same three bets.
Frequently asked questions
Why do people reply to an event dinner invite and then not show up?
Because a reply is a low-cost signal of interest and attendance is a high-cost commitment made against a conference schedule that fills up fast. The gap between the two is normal and it is not a copy problem. It becomes a planning problem only when the reply count is treated as a headcount, which is how rooms get booked for twice the number of people who arrive.
How do you measure which channel filled a side event?
Give each channel its own registration link, record the channel on the RSVP itself, and keep a door list on the night. Three separate records, because each one measures a different conversion. Without the door list you will never know which channel produced attendance rather than interest, and the door list is the one of the three that usually goes unkept.
How far ahead should event dinner invitations go out?
Far enough that the recipient’s conference calendar is still open, which in practice means several weeks before the event rather than the week of it. Senior attendees at a large conference commit their evenings early, and an invitation that arrives once those evenings are gone is competing for a slot that no longer exists, regardless of how good the invitation is.
Is LinkedIn or email better for filling a side event?
It depends on which one can run uninterrupted for your team and which one you can measure. Social outreach often generates more replies. Email more often generates attendance, partly because a calendar-ready confirmation is easier to send and partly because email campaigns are less likely to be paused mid-flight. Run both, measure them separately, and do not assume last time’s answer holds in a different region.
How many invitations do you need to fill an event dinner?
Work backwards from three separate conversion rates rather than one. Invitations to replies, replies to RSVPs, and RSVPs to people at the door. Each is a real drop and each varies by channel. A single blended assumption produces either an empty room or a wasted deposit, and the only way to get the three numbers is to have measured your own last time.
