Your registration data from last year is the only forecast worth having

Averages describe a day nobody attends

The standard planning mistake is to divide delegates by the length of the registration window and size the desk against that number. A 2,400-delegate event with a three-hour window becomes 800 an hour, which sounds manageable.

Nobody arrives at the average rate. Arrivals cluster, and they cluster for reasons that are entirely predictable once you look: the published start time, the coach schedule, the metro timetable, and whether breakfast is included.

What the curve usually looks like

For an Indian corporate conference with a 09:30 keynote and doors at 08:00, the shape we see repeatedly is:

• //A thin trickle from 07:45 — early arrivals, usually speakers and sponsors
• //A sharp rise from 08:35
• //Peak between 08:50 and 09:15, carrying 35 to 45% of the total
• //A long tail after the keynote starts, at 5 to 10% an hour

The peak half hour, not the three-hour window, is what your desk has to survive. Size against the window and you have designed a queue without noticing.

Pull the timestamps you already have

If last year used any digital registration or access system, it recorded a scan time against every delegate. That file is your forecast. It beats every heuristic in this article, including the numbers above.

What to extract:

Arrivals per five-minute bucket, across the whole morning. Five minutes, not fifteen — a fifteen-minute bucket smooths away exactly the spike you are trying to see.

The busiest thirty consecutive minutes, as a share of the day's total. This single number drives your counter count.

Arrivals before doors officially opened. If a meaningful number scanned before the published time, your desk needs to be live earlier than you planned, which is a staffing and catering decision, not a technology one.

What changes the curve year on year

Same venue, same start time, same delegate mix: the curve is stable to within a few percent. Trust it.

Change any of these and re-model:

• //Venue transport. A venue on a metro line spreads arrivals; one reached only by coach concentrates them into drop-off windows.
• //A paid breakfast becoming free. This pulls a surprising share forward by twenty minutes.
• //A keynote by someone people actually want to see. Late arrivals collapse; the peak sharpens.
• //Pre-event badge mailing. Removes those delegates from the desk entirely, which flattens everything.

Turning the curve into a decision

Once you have peak arrivals per hour, the arithmetic is short. Counters needed equals peak arrivals per hour divided by throughput per counter per hour, rounded up, plus one for failure.

At nine seconds a delegate a counter clears 400 an hour. A peak of 1,600 an hour therefore needs four counters, plus one, which is five.

The value of the real curve is that it turns a negotiation about how many counters feel right into an arithmetic problem with an answer. That argument is much easier to win, and much easier to be held to afterwards.