10,155 of 21,585 Orders Went Through Kiosks — 90 Days of Shrinking the Counter

The short version

Ticketing throughput isn't set by how many counters you have open. It's set by where people stand, what they look at, and how many steps they take. Across 13 events run on Ticspass between July 15 and October 13, 2026, we tallied 21,585 paid orders by channel: 10,155 through kiosks, 3,178 through on-site POS. Looking only at the two on-site payment paths, kiosks accounted for 76.2% and POS for 23.8%. Machines took roughly 3.2 times as many orders as people did.

But the ratio itself isn't the point. The layout that produced it is. Installing a kiosk doesn't get you to 76% on its own.

We opened another counter, and the line stayed exactly the same

When the line gets long on site, the first instinct is always to add a counter. Assign one more staff member, set up one more desk. Yet adding counters often fails to shorten the perceived wait, because the bottleneck isn't at the point of payment — it's upstream of it.

A single on-site ticket sale involves far more than a payment. You ask which ticket type, check whether a discount applies, count the party, hand over the receipt and the tickets, and point them toward the entrance. Whether you have two counters or three, that conversation happens once per customer. Adding counters increases how many conversations run in parallel; it doesn't make any single conversation shorter.

Worse, adding counters adds lines. In a lobby with a fixed width, going from two lines to three means those lines start cutting across the entry path and the exit path. From that moment on, the venue is jammed regardless of how fast you can sell tickets.

90 days of measurement — order distribution by channel

The measurement window runs 90 days, from July 15 to October 13, 2026, and covers all 13 events on Ticspass, aggregated anonymously. There were 21,585 paid orders; the distribution of orders with a recorded sales channel is below.

Three things to read here. First, most on-site payments have already moved to machines. Second, POS still handled 3,178 orders anyway. Third, there's a separate stream of orders arriving from outside the venue — web, reservation form, payment module route. These three blocks don't substitute for one another; each does a different job.

The orders kiosks take and the orders people take are different orders

Orders that migrate to kiosks share a profile. Simple ticket type, small party, card payment. Pick a ticket on screen, insert the card, done. This type has no reason to involve a conversation.

The 3,178 orders that stay at POS are a different animal: group headcount adjustments, verifying discount eligibility, cash payments, cancellations and changes, foreign-language assistance. Push these onto machines and the whole system slows down, because you've built a structure where someone deliberates in front of a screen with a line forming behind them.

So the design goal is not "100% kiosk." It's to shape the entry path so that simple orders go to machines and exception orders go to people, automatically. When that split works, each POS counter handles fewer transactions — but gains more time to spend on each one. That's the real reason the line in front of the counter gets shorter.

What we shrank, and what we widened

When we redrew the flow, here's what we actually changed.

1. Move kiosks out from beside the counter and into the first position visitors encounter at the entrance. Put them next to the counter and people reflexively join the human line.
2. Leave enough room in front of the kiosks for the line to grow in one direction only. If the machines are too close together, two units get used like one.
3. Reduce the number of counters and convert them into dedicated exception handling, then clear the floor space in front of them.
4. Separate the exit direction for people who have finished paying and collecting tickets so it doesn't overlap the entry queue.
5. Route online ticket holders straight to the entry point without passing through the ticketing line.

Drop items 4 and 5 and the first three lose half their effect. However fast the machines are, if the person who finished paying can't get out, their spot blocks the next person's spot. Ticketing throughput is ultimately the sum of "how fast one person vacates their position."

Online channels don't replace the box office

Over the same period, web orders came to 1,116, the reservation form to 105, the payment module route to 224, and the app to 13. By volume alone, that's small next to on-site payments. But these orders behave differently. They're already paid, so the time they consume on site is verification only — not payment.

In other words, each online order removes a slot from the on-site queue in advance. That's why, the lower an event's advance-sales share, the more kiosk placement matters; and the higher the advance-sales share, the more the width of the entry verification point matters. The two cases call for fixes in different places.

Ticspass runs at venues with very different characters — Lee Ungno Museum, Daejeon Goam Art and Culture Foundation, BILIE, SETEC, among others. The steady flow of museum visiting hours and the post-opening surge at an exhibition hall produce entirely different waiting curves. That's why the same kiosk gets a different placement and different operating hours depending on the site.

What changed

What these 90 days confirmed is simple. Machines can take most on-site payments, and people get faster when they focus on the orders machines can't take. Changing the shape of the line beats adding counters on actual throughput.

And you don't have to make this call on instinct. Order counts by channel accumulate every day. Watch which way the kiosk-to-POS ratio is tilting, or whether POS traffic spikes only in certain time windows, and you'll know where to widen next. Change the layout, and the next day's numbers give you the answer.

Frequently asked questions

Can we cut box office staff if we install kiosks?

The goal isn't eliminating staff — it's dividing roles. Even across these 90 days, on-site POS handled 3,178 orders. Orders that require human judgment — groups, discount verification, cash, changes — don't go away. The realistic structure is to keep a counter dedicated to those orders while sending simple transactions to machines, so each counter transaction gets more time.

How many kiosks should we install?

Placement comes before unit count. If the machines are too close together, two units merge into a single line and get used like one. Check first whether the kiosks are the first thing visitors meet at the entrance, whether there's room for the line to grow in one direction, and whether people who finish paying can exit along a path separated from the entry queue. Then decide the count.

If online sales grow, does the on-site box office become unnecessary?

Over the same period we took in 1,116 web orders, 105 reservation form orders, and 224 through the payment module route — yet on-site payments kept coming, 10,155 via kiosk and 3,178 via POS. Online orders remove a slot from the on-site queue in advance, and to that extent the width of the entry verification point becomes the thing that matters on site. It's a division of labor, not a replacement.