Skip to content

Show opening surge management using last edition door data

Attendee analyticsUpdated 2026-08-188 min read

In short

Show opening surge management sizes entry capacity against the busiest quarter hour rather than the morning average. On a 9,840 arrival day with 38 per cent through the doors before 10:30, the mean rate is 41 people a minute while the 09:00 bin runs at 70, and the gap between those two numbers is the queue photograph.

The photographs that circulate after a show are almost always taken between 09:00 and 09:45 on day one. Show opening surge management is the work that decides whether those photographs exist, and it is mostly done six weeks earlier with an export of last edition's door timestamps and a spreadsheet.

The thing that goes wrong is nearly always the same. Somebody sized the entrance against a rate that averages the whole morning, and the morning does not arrive at its average.

What the first forty-five minutes actually decide

Three things get decided at the doors, and only one of them is about attendance.

The first is the experience an attendee has before they see a single stand, which sets the tone for their day and shows up later in survey scores that get attributed to the content programme. The second is where your staff are for the rest of the morning, because a large queue pulls people off the floor and they do not go back. The third is the safety position, since a dense crowd converging on a narrow opening is the only genuinely hazardous state most exhibitions produce.

Everything else about the day is recoverable. A queue at 09:15 is not, because by the time you notice it the people are already there.

Reading the surge out of last edition's data

The input is entry scans or door counts with timestamps, binned into quarter hours. Constructing that arrival curve properly, including what to do about doors you did not instrument, is a job in itself and has its own treatment. For surge planning you need three things out of it.

The total in the surge window. The shape inside that window. And the variation in that shape across previous editions.

Take a day with 9,840 arrivals. Suppose 38 per cent of them come through before 10:30, with doors opening at 09:00. That is 3,739 people in 90 minutes, which averages to 41.5 a minute.

Hold on to that 41.5, because it is the number that ruins entrances. It is a true statement about the window and a bad basis for sizing anything.

Why does an average arrival rate under-size the doors?

Bin the same 3,739 people into the six quarter hours between 09:00 and 10:30 and the shape appears.

The 09:00 bin takes 1,050. The 09:15 bin takes 890. Then 620, then 480, then 380, and the 10:15 bin takes 319. Those six add to 3,739.

The mean bin holds 623 people. The peak bin holds 1,050, so the peak-to-mean ratio is 1.68. Converted to rates, the peak bin runs at 1,050 over 15, which is 70 people a minute, against the 41.5 a minute the window average suggested.

Now size the lanes. Suppose an entry lane with a badge scan and a turnstile clears 18 people a minute, which is a person every 3.3 seconds and is a reasonable figure for a well-run gate with staff beside it.

Against the average of 41.5 a minute you need 41.5 over 18, which is 2.3, so 3 lanes. Against the peak of 70 a minute you need 70 over 18, which is 3.9, so 4 lanes.

Open 3 lanes and the peak bin arrives at 70 a minute into a capacity of 54 a minute. The deficit is 16 a minute, and across the fifteen minutes of that bin, 240 people accumulate in front of the doors. They are still there at 09:15, when the next bin arrives at 59 a minute into the same 54, and the queue keeps growing until 09:30.

One lane, costing one turnstile and two staff, was the entire difference. And nothing in the average rate of 41.5 told you it was missing.

Size on the ninetieth percentile bin

The 1,050 figure has a problem of its own. It came from one edition, and next edition will be different in ways you cannot forecast, since the peak bin depends on train times, on weather, on whether the opening keynote starts at 09:30 and on what the show next door is doing.

The fix is to treat the peak bin as a distribution rather than a number. Pull the busiest quarter hour from each of the last five editions and look at the spread. Suppose they come out at 880, 940, 1,050, 1,120 and 1,190.

The mean of those is 1,036, which is close to the figure you already had. The 90th percentile sits around 1,162, which is 77.5 people a minute, which needs 77.5 over 18, so 4.3 lanes, so 5.

Five lanes rather than the four you would have specified, and rather than the three the window average implied. The extra lane is insurance priced against your own history rather than against a supplier's assurance, and the argument for it is a table of five numbers that any finance lead can read.

I would size on the 90th percentile and not the maximum. Sizing on the worst edition you ever had builds an entrance that stands idle for 358 days and still fails on the 359th, because there is always a worse edition available. The percentile makes the risk explicit: you are accepting that roughly one edition in ten will exceed the design, and you plan an overflow response for that case rather than paying for it in steel.

The queue that already exists when the doors open

Here is the part that the arrival curve conceals, and it is the reason surge management is a different job from door sizing.

Those 1,050 people in the 09:00 bin did not arrive between 09:00 and 09:15. A large share of them arrived between 08:30 and 09:00 and stood outside waiting. If 620 people are already in front of the doors at 08:58, then at 09:00 the instantaneous demand is 620 people simultaneously, which no lane count in the world absorbs, followed by a much gentler flow from people still arriving.

Your entrance therefore faces two quite different problems ten minutes apart. The release of an accumulated crowd, and then a genuine arrival stream. Lane capacity solves the second. Only staging solves the first.

Staging means a holding area with a defined boundary, a defined capacity and a defined release. People are admitted into the holding area before the doors open, which converts a queue in the weather into a queue under cover, and they are released into the lanes at a rate the lanes can take. In practice that means opening the doors in a controlled way, with stewards metering the flow, rather than removing a barrier and stepping back.

It also means opening earlier where you can. Every person you process at 08:40 is a person absent from the 09:00 release, and shifting 300 people out of that bin is worth more than a fourth lane. If your badge collection can run before the halls open, run it, because badge collection and hall entry are separable problems and combining them puts both on the same fifteen minutes. Where check-in itself is the constraint, the desk arithmetic in registration desk throughput tells you whether the doors or the desks are the binding limit.

Why does a door surge block itself?

The reason metering works, and the reason a wide-open door underperforms its theoretical capacity, lies in how crowds behave on the approach.

Helbing and Molnar published a model in Physical Review E in 1995 describing pedestrian motion as governed by social forces: an acceleration towards a desired speed and direction, a repulsion from other people and from boundaries, and attraction terms. The behaviour that falls out of it is the one every operations lead recognises. A crowd converging on a narrow opening does not form an orderly funnel. It forms a pressing arc, and people at the edges of that arc obstruct the people who would otherwise be moving through the middle.

Helbing, Farkas and Vicsek took this further in Nature in 2000, examining the mechanisms of and the preconditions for panic and jamming produced by uncoordinated motion in crowds, and drawing practical conclusions about reducing the harm.

The operational reading is straightforward. Flow through an opening depends on how people approach it, so your measured lane rate of 18 a minute assumes an orderly feed and will not survive a disorderly one. Barriers that form an orderly queue before the lanes, instead of letting a mass gather in front of them, are therefore doing capacity work as well as tidiness work, and the same is true of the stewards metering the release.

Where this stops

The whole method rests on last edition's shape holding, and there are four situations where it plainly does not. A changed opening time. A changed venue or a changed entrance. A keynote or ceremony that starts at a fixed time and pulls the whole distribution forward. And a large co-located event sharing your doors.

In any of those cases the historical bins are actively misleading, and the honest move is to widen the design margin rather than to model harder. Take the 90th percentile figure and add capacity for the specific mechanism you can name.

The second limit is that this is a plan, and a plan is not a response. You still need somebody watching the actual arrival rate against the plan in the 09:00 bin, with the authority to open a lane or open a second door without asking permission. The numbers above tell that person what normal looks like, which is the useful half of the exercise.

There is a third limit worth stating plainly. Everything here treats attendees as an undifferentiated flow, and they are not. VIP lanes, exhibitor entrances, group arrivals from coaches and press all behave differently and often share doors. A model that gets the aggregate right and puts a 60 person coach group into a lane sized for individuals will still produce the photograph.

The step for this week is one query and five numbers. Pull the busiest quarter hour of entries from each of your last five editions of one show, sort them, and take the fourth. That is roughly your 90th percentile bin. Divide it by 15, divide that by your lane rate, and compare the answer with the number of lanes you actually opened. If you were short, you now know by how much, and everything else in your on-site attendee analytics can wait behind that. If the arrival rate turns out to be fine and the desks behind the doors are the constraint, the fix belongs in a registration desk staffing model, and if the kiosks are the constraint then self service check in throughput is where the missing capacity is hiding.

Questions people ask about show opening surge management

How do you plan for the opening rush at a trade show?
Take last edition's entry timestamps, bin them into quarter hours, and find the busiest bin rather than the hourly average. Size your entry lanes against that bin, allow for the queue that has already formed before the doors open, and stage the release so the accumulated crowd enters at a rate the lanes can absorb.
What percentage of attendees arrive in the first hour?
It varies by sector and by whether a keynote opens the show, so use your own figure rather than a benchmark. A profile where 38 per cent of a 9,840 arrival day passes the doors in the first 90 minutes gives 3,739 people, or about 41 a minute averaged across that window, with the true peak considerably higher.
Why does an entrance jam even when it has enough capacity?
Because people arriving at a doorway push towards it faster than it can absorb them. Helbing and Molnar modelled pedestrian motion as subject to social forces in 1995, and Helbing, Farkas and Vicsek examined jamming produced by uncoordinated motion in 2000. A crowd converging on a narrow opening obstructs itself before any capacity limit is reached.

Related reading

All on-site analytics articles