The line at 8:05 is not a printer problem. It is a peak-hour sizing problem. By the time the queue wraps past the coffee station, the arithmetic that produced it was already fixed — not by the number of printers on the loading dock, but by the assumption that arrivals spread evenly across the registration window. They do not. They concentrate, and the concentration is sharpest at exactly the moment when a keynote is about to start and every attendee in the building wants a badge at once.
This article is about sizing the entrance to the busiest hour rather than the guest list, separating the exception lane from the fast lanes, and treating printer choice as a queue decision. The method is arithmetic. The inputs are yours. The published figures below are a starting range to be corrected against your own arrival data, not a promise.
Size for the peak hour, not the guest list
The instinct is to scale the entrance to total attendance. Two thousand people, so a big desk. It is the wrong input, because those two thousand people do not arrive evenly across the registration window. They arrive in a surge. Published planning figures put 60 to 80 per cent of arrivals inside a window of 60 to 90 minutes, and the pattern is sharpest exactly where it hurts most, at morning conference openings tied to a keynote (Converve).
Your entrance does not have to process two thousand people. It has to process roughly fourteen hundred of them in ninety minutes, and then stand nearly idle for the rest of the day. That asymmetry is the whole planning problem. Size for the average, queue at the peak.
Two consequences follow for the person signing the hardware order. First, the working number in every calculation below is arrivals in the peak hour, never total registrations. Second, the assumption behind it belongs in writing. If you are planning on seventy per cent of confirmed registrations turning up inside ninety minutes, write it down, then measure the real curve on the day and correct it next time. An assumption nobody wrote down cannot be proved wrong, which is what makes it dangerous.
The published spread is wider than you think
Before you can size anything, you need a per-person processing time. This is where the published guidance stops agreeing with itself. One vendor guide puts a staffed check-in desk at 40 to 60 people an hour. Another puts it at 60 to 80. A third works with 120. All three are describing the same desk (Converve).
A factor of three between the low and the high figure is not a rounding difference. It is the difference between four stations and twelve, and no published guide explains where the spread comes from. It comes from four variables that vendor content tends to average away:
- Whether your staff search by name or scan a code.
- Whether the badge already exists or has to be printed.
- Whether exceptions queue in the same line as everyone else.
- How many people arrive at once.
Get those four right and the station count falls out of the arithmetic. Get them wrong and you can buy the best check-in software on the market and still build a queue.
For a rough starting range, published throughput figures put kiosks at roughly 80 to 120 an hour per unit, against 40 to 60 for a staffed desk working from pre-printed badges (Converve). These are vendor planning figures rather than independent measurements, and they are best used as a starting range to be corrected against your own arrival data. Treat them as a shape, not a promise. What they do establish reliably is the direction of the difference: the jump from a paper list to a scan is enormous, and the jump from a scan to a kiosk is real but far smaller.
The four-step calculation
Once you have a peak-hour figure and a per-person time, the station count is arithmetic. Divide 3,600 seconds by your processing time to get the throughput of one station, then divide peak arrivals by that number, then add the exception lane. The method below is assembled from published staffing formulas (Converve). The inputs are yours; only the arithmetic is borrowed.
Step 1: Count the peak hour
Take expected arrivals in your busiest 60 minutes, not total registrations. Without last year’s data, start at 70 per cent of expected attendance across 90 minutes. Write the assumption down.
Step 2: Set a realistic per-person time
Scan and print at a staffed desk is 45 to 60 seconds once you include the greeting. A kiosk with a clean pre-registration record is under 30. A name search without a code is minutes, not seconds.
Step 3: Divide
3,600 seconds divided by your per-person time gives one station’s hourly throughput. Peak arrivals divided by that gives the number of stations you need.
Step 4: Add the exception lane
One dedicated help position per four to six fast lanes, plus reserve badge stock. This is the step that most plans skip, and the one that decides whether the fast lanes stay fast.
A worked example, clearly labeled as a hypothetical
Work it through for a conference of 2,000 with 70 per cent arriving inside 90 minutes. That is roughly 930 arrivals in the busiest hour. At 60 seconds per person a station handles 60, so you need about 16 positions. At 30 seconds each you need 8. The difference between those two answers is not the software brand on the invoice, it is whether the badge has to be searched for or scanned.
Every input in that paragraph is an assumption. The 2,000 is your registration count. The 70 per cent is a planning figure from published guidance, not a measurement of your event. The 60 and 30 seconds are ranges from vendor documentation. Replace them with your own arrival curve and your own measured per-person time as soon as you have them. The arithmetic is the only part that transfers.
Then add the lane the arithmetic does not produce. A supplier case describes a 2,000 person conference that opened with eight stations, hit congestion immediately, and recovered only after moving to fourteen stations with a serpentine queue and a separate desk for problem cases, at which point waits fell from 25 minutes to under five (Converve). The extra stations helped. Pulling the problem cases out of the fast lane helped more. A broken record blocks more than a missing station.
Printer choice is a queue decision
Hardware is where the calculation quietly gets decided, because the two common badge technologies are not in the same performance class and the specification sheets say so plainly.
A direct thermal label printer such as the Brother QL-820NWBc is rated at up to 110 labels a minute at 300 dpi, and it consumes no ink and no toner at all. The heat sensitive stock is the entire recurring cost. A plastic card printer such as the Zebra ZC300 is rated by its manufacturer at 200 single-sided colour cards an hour, against 900 an hour if you drop to monochrome. Its colour ribbon cartridge is rated for 200 images, so a full colour card programme also means swapping ribbons during your peak (Converve).
Those are rated batch speeds, not check-in throughput. Nobody prints at rated speed at a check-in desk, because the human in front of the desk is slower than the machine. The point is the ratio. A thermal badge is never the bottleneck in your queue, and a full colour plastic credential can be.
There are good reasons to print cards: multi-day access credentials, reuse, a premium delegate tier, durability in a lanyard for a week. Just choose them knowing what they do to the morning. The colour decision costs you the speed.
The exception lane is not a courtesy desk
The exception rate that lane absorbs is not bad luck. It follows from what a badge has to represent at a business event. At a consumer event a badge proves someone paid; at a trade show or a business conference it does considerably more. Your exhibitors, your hosted buyers, your trade visitors, your press, your speakers and your build-up crew hold different tickets, enter on different days, reach different areas and appear differently in the matchmaking system. The badge is the visible end of a permissions record.
When badge type and ticket category are the same underlying record, a corrected registration produces a corrected badge automatically. When they live in two systems, somebody reconciles them by hand at seven in the morning. That is also why the exception lane is not just a courtesy desk. A buyer who upgraded last week, an exhibitor who sent a substitute, a speaker added on Friday: none of these are errors, they are normal business-event traffic, and they all take longer than thirty seconds.
Supplier guidance suggests holding 10 to 15 per cent of expected attendance as pre-printed backups plus blank stock for walk-ups, and printing 50 to 100 test badges across every station the evening before (Converve). That test run is not a formality. It is how you find the station with the misaligned stock or the ribbon that will run out at 8:20.
When self-service is the wrong answer
Vendor content rarely says this, so it is worth saying plainly. Kiosks are the wrong choice in at least four situations.
- Low pre-registration: the kiosk advantage depends on people arriving with a code. If half your attendees register at the door, you are running an on-site registration desk with a kiosk parked next to it.
- Accreditation with document checks: where a trade visitor has to prove professional status or a badge has to be matched to identification, a person has to look at something. That is a staffed process by definition.
- Premium hosted formats: for a hosted buyer programme, the welcome is part of what the buyer was promised. Sending a flown-in buyer to a touchscreen saves ninety seconds and costs you the first impression you paid four figures to create.
- Complex badge types in a small crowd: below roughly 300 attendees with six ticket categories, the exception rate is high enough that the kiosk mostly generates work for the help desk.
The honest version of the advice is a mix. Kiosks for the pre-registered majority, a staffed lane for anyone whose record needs a human, and a separate desk for people registering on the day.
What the arrival curve looks like at scale
For a sense of how concentrated arrivals can be, consider a documented case. Becknology Days, a multi-day agricultural event, includes seven regional events that draw 1,000 to 1,500 attendees each, and a flagship event that welcomes over 13,000 attendees across three days. The events are 100 per cent free and 100 per cent day-of, walk-up registrations. The main event tackles up to 4,000 registrations in a single morning (Expo Pass).
That is not a pre-registered conference with a code scan. It is a walk-up registration desk at volume, and it is a useful reminder that the arrival curve is not always the one you planned for. If your event has any walk-up component, the exception lane is not an exception. It is a main lane.
Frequently asked questions
How many check-in stations do I need?
Divide expected arrivals in your peak hour by 3,600 seconds divided by your processing time per person, then add one help position per four to six lanes. For a 2,000 person conference with 930 arrivals in the peak hour, that is roughly 16 stations at 60 seconds per person, or 8 at 30 seconds.
Should I pre-print badges or print on demand?
Print on demand if you expect late registrations, name changes or walk-ups, which describes most business events. Pre-printing is defensible for a closed, confirmed list that will not change, and it still needs a printer on site for corrections.
How many spare badges should I hold?
Supplier guidance suggests holding 10 to 15 per cent of expected attendance as pre-printed backups plus blank stock for walk-ups, and printing 50 to 100 test badges across every station the evening before (Converve).
What is the difference between rated printer speed and check-in throughput?
Rated speed is the machine printing in ideal conditions. Check-in throughput is the machine plus the human scanning, greeting, printing and attaching. The human is almost always the slower component, which is why the useful comparison between printer technologies is the ratio, not the headline number.
The copyable artifact: a peak-hour sizing worksheet
For the check-in lead, here is a one-page worksheet you can fill in before doors open and re-run after the event with measured data. Every field is an input you control or measure.
PEAK-HOUR CHECK-IN SIZING WORKSHEET
1. ARRIVAL CURVE
Total expected attendance: ______
Assumed % arriving in peak 90 min: ______ (start at 70% if no prior data)
Peak-hour arrivals (working number): ______
2. PER-PERSON TIME
Scan + print + greet (staffed): ______ sec (start at 45-60)
Kiosk with clean pre-registration: ______ sec (start at under 30)
Name search without code: ______ sec (minutes, not seconds)
3. STATION COUNT
One station hourly throughput: 3600 / ______ = ______
Stations needed: Peak arrivals / throughput = ______
4. EXCEPTION LANE
Fast lanes: ______
Help positions (1 per 4-6 lanes): ______
Reserve badge stock (% of expected): ______ (start at 10-15%)
Test badges printed evening before: ______ (start at 50-100)
5. POST-EVENT CORRECTION
Measured peak-hour arrivals: ______
Measured per-person time: ______
Measured exception rate: ______
Revised station count for next time: ______
The worksheet is reversible. Run it with your assumptions before the event, then run it again with measured data after. The gap between the two is your planning error, and it is the most useful number you will produce all year. The line at 8:05 is not inevitable. It is the visible output of an assumption nobody wrote down.