An event's no-show rate is the share of registered seats that never checked in: (registered − checked in) ÷ registered. Measure it from your own door list after every event, split by paid and unpaid seats, and within a season you'll have a show rate to multiply against next month's registrations. That number sets the catering order, the door staffing, and how far, if at all, you can sell past the room.
The formula, and what counts as a no-show
Count seats, not registrations: a family of five on one registration is five seats. A cancellation before the door opens comes out of registered, and a refund is a cancellation. A walk-up who never registered isn't in the formula; count walk-ups separately.
The complement is the show rate, checked in ÷ registered, and that's the number you'll carry forward. A 12% no-show rate is an 88% show rate. On one event it's a fact about that night. Across a season, split by ticket type, it's a forecast.
Measure it from your own door list
Start with the last event that already happened. You need seats registered when the door opened and seats checked in when it closed, per event and per ticket type. Paid and unpaid is enough to start.
- Seats registered
- 150 (the sample season's sold count; use yours)
- Checked in at the dock
- 131 (yours comes off the door list)
- No-shows
- 150 − 131 = 19 seats
- No-show rate
- 19 ÷ 150 = 12.7%
- Show rate
- 131 ÷ 150 = 87.3%
One sailing is a data point, not a rate. Log it; the rate is what the log says after eight or ten of them.
Splitting by segment is where the number earns its keep. The registration list in the free app carries a quantity, a paid flag, and a check-in flag on every entry, so a list filtered to one event is the whole input. Here's the sample list for the Friday Sunset Cruise, read as if the boat had sailed.
- Registrations
- 5, covering 14 seats
- Paid seats
- Kelly Hammel 2, Marcus Bell 4, Priya Raman 2, Dana Whitfield 1 = 9
- Unpaid seats
- The Okonkwo family, 5
- Checked in
- Kelly Hammel and Marcus Bell, 6 seats, all paid
- Paid no-show rate
- (9 − 6) ÷ 9 = 33%
- Unpaid no-show rate
- (5 − 0) ÷ 5 = 100%
- Overall no-show rate
- (14 − 6) ÷ 14 = 57%
Fourteen seats is far too few to trust: one family of five moved the overall rate by 36 points. Keep the split on every event, then pool each segment across events until it holds a few hundred seats.
Benchmarks: free events vs paid events
Published figures are a sanity check for your first season, not a substitute for your log. Two things hold across every source below: free events lose far more registrants than paid ones, and the gap between a good night and a bad one is wide enough that an average alone will mislead you.
| Source | What was measured | Figure |
|---|---|---|
| Pheedloop, Event Data Lab Report #05 (April 2026) | Median no-show, 1,070+ in-person events with check-in data | About 20% overall; about 28% for free events (389 events), about 17% for paid (683 events) |
| Same report, the bad nights | 25th percentile, and events losing over half their expected attendance | 47% no-show for free events vs 32% for paid; 22% of free events lose over half, vs 14% of paid |
| Same report, by size | Median no-show by attendee count | About 32% at 10 to 49 attendees, 23% at 50 to 149, 19% at 150 to 499, 18 to 19% above 500 |
| Eventbrite UK organizer roundup (April 2021) | Organizers' own dropout figures for free events | Quoted rates run 30% to 50%; one plans for 50% dropout on every free event. One over-allocated 50 to 60 tickets on a 400-person target and landed near 85% attendance |
| Livestorm, 2026 Webinar Benchmark Report | Show-up across 33,786 webinar sessions and 7.06 million registrations in 2025 | 51.3% average show-up, a 48.7% no-show rate for mostly free online sessions |
What isn't in that table is a paid no-show rate by ticket price. None of these sources splits paid events by price band, so the gap between a $35 class and a $175 tournament is something only your own log can show. The free-versus-paid gap is about commitment, so treat a registered-but-unpaid seat as its own segment even at the same price; it behaves like a free seat until the money clears. If you run free events and the no-shows hurt, the lever is a small commitment at registration, a refundable deposit or a nominal ticket, with the show rate logged before and after.
Forecasting attendance from registrations
Once the log has a show rate per segment, the forecast is a weighted sum: split current registrations into the segments you measured, multiply each by its historical show rate, and add them up.
Use your own pooled show rate as soon as you have one. Until then, borrow the benchmark medians above and label the forecast as borrowed.
- Seats sold so far
- 142 of 200 (sample season)
- Paid seats
- 134 (your split; the sample list's teams pay when they register)
- Registered, unpaid
- 8 (your split; J. Reinhart's 2 unpaid seats are among them)
- Paid show rate from your log
- 94% (your number)
- Unpaid show rate from your log
- 60% (your number)
- Expected attendance
- 134 × 0.94 + 8 × 0.60 = 126 + 4.8, about 131 shooters
Hold it loosely this far out. Registrations arrive late: Livestorm's 2026 report found 49.6% of webinar registrations land in the final week and 15.3% on the day itself. It's webinar data, but the shape is the point: re-run the forecast weekly.
Using the forecast: catering, staffing, and the oversell decision
- Catering: order to the forecast, then round toward the cheaper mistake. On a dinner cruise, running out costs more, so round up; where leftovers do, round down.
- Door staffing: time one check-in at your own door, multiply by expected attendance, and divide by the minutes you'll let people queue. Staff to expected attendance, not registrations.
- Waitlist releases on a sold-out event: release a seat when a cancellation comes in or a no-show is confirmed at the door, unless you're deliberately overselling.
Should you oversell to cover no-shows?
Sometimes, and the rule is short. Oversell only when the capacity is a preference rather than a legal or safety limit, when turning someone away costs less than an empty seat, and when your log is long enough to show your best-attended night. Size the oversell to the highest show rate you've recorded, so even a full night fits.
A vessel's certificate of inspection, a fire marshal's occupancy number, or a count of shooting stations is a hard ceiling. Don't oversell past it; run a same-day standby line and fill confirmed no-shows from it.
- Capacity
- 120 seats, a vessel limit (sample season)
- Paid show rate from your log
- 87.3% median, 96% on your best night (your numbers)
- Expected empty seats at the median
- 120 × (1 − 0.873), about 15
- Arithmetic oversell ceiling
- 120 ÷ 0.96 − 120 = 5 seats
- What to actually do
- Nothing past 120. The certificate is the cap, so those 15 likely-empty seats go to a standby line at the dock
A free event in a room with headroom runs the same arithmetic against a softer ceiling; the Eventbrite UK over-allocation in the table above is that case.
A reminder sequence that lowers the rate
Reminders don't change who committed; they change who forgot. Treat the sequence below as a schedule to test: run it for a season, compare each segment's show rate before and after, and keep the touches that moved it. Published effect sizes come mostly from medical scheduling, so measure your own.
| Touch | When | Channel | What it carries |
|---|---|---|---|
| Confirmation | On registration | Date, time, address, what to bring, an add-to-calendar link, and how to cancel. A cancellation is a seat you can resell; a no-show isn't. | |
| One week out | 7 days before | Details again plus anything new. Ask unpaid registrants to pay now or release the seat. | |
| Day before | 24 hours before | Text with consent, otherwise email | Short: time, address, what to do if they can't make it. |
| Morning of | 2 to 4 hours before | Text | One line: doors open at this time, reply if you're running late. Replies feed the standby line. |
Track it every event
- Close the door list after each event and record seats registered, seats paid, seats checked in, and paid seats checked in. Seats, not registrations.
- Compute the no-show rate overall and per segment, and note free or paid, ticket price, and capacity so you can sort the log later.
- After eight to ten events, pool each segment and take its median show rate and its highest. The median feeds the forecast; the maximum sets the oversell ceiling.
- Before each upcoming event, split current registrations into the same segments, multiply by the segment show rates, and set catering, staffing, and standby counts from the result.
- Keep the log in the no-show log template below, or start in the registration list: enter each registration with its quantity and paid status, check people in at the door, and read registered and checked in off the event summary.
Questions people ask
What is a good no-show rate for an event?
Lower than your own last season. For a sanity check, Pheedloop's April 2026 analysis of 1,070+ in-person events put the median at about 20%, paid events near 17%, free events near 28%, and small events of 10 to 49 people around 32%. A free event near 50% is where several organizers in Eventbrite UK's 2021 roundup said they plan to be: a reason to add a commitment step, not to panic.
Should you oversell an event to cover no-shows?
Only when the capacity is a comfort limit rather than a legal or safety one, when a turned-away guest costs less than an empty seat, and when you've logged enough events to know your best-attended night. Size the oversell to that night's show rate, never the average. Where a certificate of inspection or an occupancy number sets the cap, don't oversell; run a standby line instead.
How do you count a no-show when one registration covers several seats?
Count seats. A registration for four that arrives as two people is two no-shows, not zero and not one. That's why the log tracks seats registered and seats checked in rather than names, and why the registration list carries a quantity on every entry.