Crowd management does not begin at the turnstiles. It begins much earlier, when someone decides where the turnstiles should go.

At seven in the morning, the stadium is empty.

The barriers are already in place, the access points have been chosen and the lanes have a set width. The coach parking areas are where someone decided they should be months earlier. The public transport stops are fixed, the exits cannot be moved and the waiting areas already have defined dimensions.

In a few hours, fifty thousand people will arrive. That is when everyone will start talking about crowd management.

In reality, much of the work has already been done. Well or badly.

This is the paradox: by the time we can see the crowd, many of the decisions that will shape its behaviour have already become difficult, if not impossible, to change.

We manage conditions, not just people.

When people talk about crowds, the first image that usually comes to mind is the steward: someone in a vest directing people, staffing an access point or regulating a flow.

This is the visible part of the system. It is not necessarily the most important.

A crowd moves through a space someone has designed, using information someone has chosen to provide, along routes selected in advance and towards destinations placed in specific locations.

If ten thousand people arrive at four turnstiles at the same time, the staff on the ground can certainly handle the situation well or badly. But before asking how they are working, perhaps we should ask another question: why did ten thousand people arrive at four turnstiles at the same time?

The problem may not have started there. It may have started months earlier.

Capacity is more than a number

ISO 22353:2026, Security and resilience — Guidelines for crowd management, was published in August 2026.

The interesting aspect is how it approaches the problem, as well as the guidance it provides. Crowd management is considered across the event’s entire lifecycle: audience characteristics and expected behaviour, space capacity, entry, internal circulation, exit, emergencies, stakeholder involvement and subsequent evaluation.

This leads to an observation that sounds obvious but is not always reflected in practice: the capacity of a place is not simply the maximum number of people we can physically fit inside it.

A space may accommodate 20,000 people. But can we get 20,000 people into it? How quickly, through how many access points and with which checks? Once inside, can they move? How long will it take them to leave? Above all, what happens if part of the system suddenly stops working?

Nominal capacity is a photograph. Operational capacity is a film.

It is the film that matters to security.

The bottleneck moves

Imagine an event with eight access points. One stops working, and the natural response is to concentrate resources on the other seven.

That makes sense. But what happens next?

If we increase throughput at security checks, we may simply move congestion a hundred metres further along: to the turnstiles, ticket checks or entrances to the stands. Every time we resolve a bottleneck, we should ask a simple question:

Where am I moving the problem?

It sounds obvious. In practice, it is anything but.

At major events, different organisations work simultaneously, each responsible for a segment of the same journey: transport, police, private security, venue management, ticketing, stewarding, the organiser and emergency services.

Each can do its own job perfectly and unintentionally make the system as a whole work less well. This is a typical risk in complex systems: optimising individual parts without optimising the system.

The public does not know our plan

There is an easy mistake to make during planning. We look at a site plan, draw arrows and define Entrance A, Entrance B, the pedestrian route, waiting area and exit.

Everything works on paper.

There is just one small problem: the public did not attend the meeting.

People arriving at an event follow what they can see. They may follow the crowd, Google Maps, a friend who has “been here before” or the first sign they can understand. A light, barrier or member of staff may help them find their way. Or they may simply follow other people.

Designing a flow therefore means more than drawing a route on a site plan. It means trying to understand which route will feel natural to someone arriving for the first time.

These are very different tasks.

There is already a crowd before the entrance

For a long time, our thinking about event security focused mainly on the perimeter: inside and outside, controlled and uncontrolled, accredited and unaccredited.

But a major event creates concentrations of people long before the checkpoint: at railway stations, underground stops, car parks, on shuttle buses, along pedestrian routes, in queues and at meeting points.

At the end of the event, the reverse happens: thousands of people leave an organised environment at the same time and enter the surrounding urban space.

Formal responsibility may change as someone crosses a road.

Risk does not.

Risk does not recognise the boundary drawn on our site plan. This is one reason crowd management should be considered as a system, rather than an isolated function.

Security and safety meet in the crowd

A suspicious person, a developing fire, a false alarm, a power outage, an unattended object, a sudden storm or a rumour spreading on social media are very different problems.

Yet all can produce the same consequence: thousands of people changing their behaviour at the same time.

This is where the organisational distinction between security and safety becomes much less clear. The origin of the event may sit within one function; its consequences affect everyone.

A crowd management plan that works only when everything goes as expected is therefore not a good plan. It must also work when something disrupts the pattern.

Leaving is part of the event

There is another phase we often continue to underestimate: egress.

The show ends and a natural pressure to leave develops. Everyone wants to go home at roughly the same time. Staff are tired, some services begin reducing personnel, public transport suddenly receives thousands of passengers, and the roads must accommodate pedestrians, taxis, coaches, private cars and the organiser’s vehicles together.

This is precisely when another problem may arise: a cancelled train, a temporarily closed station, a blocked road or an accident.

The end of the show is therefore not the end of the event.

From a security perspective, the event ends when the system has absorbed the crowd we created.

Plan for failure too

This, in my view, brings us to the central point.

A good plan should describe how the event will work and what we will do when something does not work.

What happens if we lose an entrance or exit route? If a station becomes unavailable? If access control slows by thirty per cent? If ten thousand people arrive half an hour earlier than expected? If we have to suspend entry temporarily?

This is not pessimism. It is planning.

In complex systems, the unexpected is not necessarily exceptional.

It is a design condition.

The control room comes later

We can have excellent cameras, dashboards, heat maps, people counters, flow analysis, artificial intelligence and a perfectly organised control room. They are all valuable tools.

But there is a limit no technology can remove: by the time the control room sees congestion, the congestion already exists.

Technology can help us detect it earlier. Staff can prevent it from getting worse. We can change flows, open alternative routes and change the information given to the public.

But no control room can widen a road. No algorithm can move a turnstile. No camera can create an exit we did not plan for.

This is why some of the best crowd management technology remains invisible.

It is a decision taken six months earlier.

The crowd begins while the space is still empty

Perhaps we simply need to change our perspective.

When we enter an empty stadium, a fan zone that has not opened yet or a square where the stage has not been assembled, we see a space.

We should try to see the people who will move through it: where they will slow down, where they will stop, what they will look at, where they will seek information, where a queue will form and where two flows will cross. Above all, where they will go when something stops working.

That is crowd management.

It means creating conditions that, as far as possible, reduce the need to control the crowd.

When fifty thousand people are already outside our gates, there is still a great deal we can do. But we made our best decisions when nobody was there yet.

Perhaps this is the hardest part of our work:

Seeing the crowd before it arrives.

Essential sources

• ISO, ISO 22353:2026 — Security and resilience — Guidelines for crowd management, August 2026.
• ISO/TC 292, ISO/CD TR 22356 — Security and resilience — Case studies in how effective crowd management in events, document under development.

All articles →

The Strathego Method →

Author: Francesco di Castri

Topics: AI · Crowd Management · Security