What the early bird registration deadline actually moves, and what it costs

The daily registration chart for the week of an early bird registration deadline is the most flattering picture your reporting produces all year. A flat few months, then a wall. Somebody screenshots it for the Monday commercial call, and the caption writes itself: the campaign is working.

The following Monday nobody screenshots anything, because the chart has fallen through the floor and the honest reading of the two charts together is a different story from the reading of either one.

Draw the daily curve before you draw any conclusion

Weekly totals hide the shape. Pull registrations by day for the six weeks around the cutoff and look at the daily numbers, because the deadline effect is a two or three day phenomenon sitting inside a week that otherwise looks ordinary.

A typical shape, in weekly buckets, with the cutoff at the end of week zero:

Week minus 4: 520. Week minus 3: 545. Week minus 2: 610. Week zero, the seven days ending at the cutoff: 1,400. Week plus 1: 200. Week plus 2: 260.

The spike is 1,400 and everyone can see it. The number nobody puts on the slide is 460, which is what the following fortnight produced.

The baseline decides the answer

A spike is only a spike against something. Three baselines are in common use and they give materially different answers.

Comparing the spike week to the immediately preceding week gives 1,400 against 610, a 130 per cent lift. Comparing it to the trailing four week average of 558 gives a 151 per cent lift. Comparing it to the same days out in the prior edition, which is the only baseline that controls for where you are in the campaign, gives whatever the prior edition's deadline week produced, which for most shows is another spike, so the lift collapses to almost nothing.

Use the trailing four week average for the pull-forward test below, because you want a counterfactual for what those days would have produced with no deadline in them. Use the prior edition at the same days out when you want to know whether this year's deadline outperformed last year's, which is a different question and the one your pacing report should be answering.

How much of the spike was borrowed?

The arithmetic that matters takes about five minutes and almost nobody does it.

The trailing four week average before the spike is 558 a week, and the trend was rising, so call the counterfactual for the spike week 570. Actual was 1,400, so the excess is 830.

The two weeks after the cutoff should have produced roughly 1,180 on the same rising trend, 580 then 600. They produced 460. The shortfall is 720.

So 720 of the 830 excess registrations, which is 86.7 per cent, were registrations that would have arrived anyway and arrived early instead. The deadline generated 110 registrations that the campaign would not otherwise have produced in that six week window, and it moved roughly 720 people about ten days forward in their decision.

Run this on two or three prior editions before you trust it. If the pull-forward share comes out between 80 and 95 per cent every year, that is your show's answer and you can stop re-litigating it in meetings.

The share of the file the early bird window produced

The second number worth having is structural. Take the final registration file and count how many of those registrations were created on or before the cutoff date.

On a show that closes at 11,600 registrations with 6,900 created inside the early bird window, the early bird window produced 59.5 per cent of the file. That single figure changes how you should think about the deadline entirely, because a discount applied to 59.5 per cent of your paid registrations is a pricing decision about the majority of your revenue, dressed up as a promotional tactic.

Compare that with what your audience is actually doing. PCMA reported in April 2024 on the Maritz Registration Insights Report, built from 360,000 attendee registration records across 30 trade shows over three years, that 45 per cent of 2023 conference-goers delayed registering until fewer than four weeks before the event. If nearly half your audience decides inside the final month, an early bird cutoff four or five months out is not addressing them at all. It is a price for the people who had already decided.

What does the deadline cost in planning?

The discount has a money cost, and pricing the giveaway against the people who took it is A14's arithmetic. The cost I want to argue about here is the one that never appears in a budget line.

A deadline that moves 720 registrations ten days earlier corrupts every read that depends on cumulative share at a given day out. If your forecast method is current registrations divided by the historic cumulative share at this day out, and the deadline has just parked 830 extra registrations on the curve, your day minus 120 forecast is going to be high by roughly the size of the borrowed block until the post-deadline trough works through. The forecast method itself is A9's subject. The point here is that the deadline is one of the few campaign events that systematically biases it, and it does so in the direction that makes people relax.

Two decisions commonly get taken on that inflated read. Badge and lanyard stock gets ordered against a projected total. Catering and room block guarantees get set. Both are hard to reverse and both are cheaper to get right than to fix, so the practical fix is to publish the forecast with the deadline week flagged and the pull-forward estimate subtracted, or to hold the forecast for ten days and let the trough land first.

There is also a softer cost. A deadline trains the audience. Every edition where the biggest single week is the deadline week teaches your file that registration is a deadline-driven activity, which makes the rest of the curve flatter and the show more dependent on the next deadline.

What I would actually do with the deadline

I would keep it, and I would stop treating the spike as evidence of anything.

Keeping it is defensible because a deadline is a free scheduling instrument. Pulling 720 registrations ten days earlier gives you a better read on geography and job function mix ten days sooner, and on a show where visa letters or international travel approval matter, ten days is worth something real to the attendee as well.

What I would change is the claim attached to it. The deadline does not grow the file by 830. It grows it by roughly 110 and reschedules the rest. If the internal report said that, the conversation about whether the discount is worth its cost would happen on the right numbers.

I would also stop moving the deadline around between editions. Comparing this year's deadline week to last year's only works if the deadline sat at a comparable number of days out, and shows that shuffle the date by three weeks for operational convenience have destroyed their own ability to answer the question. Announcing a date and then pushing it back does something worse than that, which is A15's subject, and how far apart the price steps themselves should sit is A13's.

A deadline is also one of the few things in the campaign that any attendee analytics report can measure cleanly, provided the daily counts are stored and not just the weekly rollups.

Freeman's 2024 Attendee Intent and Behavior study is a useful check on how much of the decision the price is carrying at all: 87 per cent of respondents named discovering new products, solutions and partners as a factor influencing attendance, 80 per cent named perceived event value relative to their professional goals, 64 per cent named overall cost, and 54 per cent named registration price. Registration price is on the list and it is near the bottom of it.

Where this stops

This measurement assumes the deadline was the only thing happening that week, and on a real show it almost never is. Deadlines get supported by an email burst, a paid retargeting push and often a sales team calling lapsed accounts. If all of that fires in the same seven days, the 830 excess is the combined effect of the deadline and the campaign spend behind it, and no amount of arithmetic on the registration file separates them after the fact.

The clean way to separate them is a holdout: withhold the deadline email from a random 10 per cent of the eligible file and compare registration rates in the two groups. Most organisers will not do this, partly for good reasons, since withholding a price warning from real customers has a fairness problem that a marketing test does not resolve. If you will not run the holdout, say in the report that the estimate attributes the whole effect to the deadline, and treat 110 net new as an upper bound.

The other limit is that pull-forward is measured over a six week window. If a registrant would have arrived in week plus five rather than week plus one, this method counts them as net new when they were merely pulled forward further. Extending the window to eight weeks tightens it, at the cost of picking up whatever else the campaign did in the meantime.

Before your next deadline, export registrations by day for the six weeks around your last two cutoffs and compute the two numbers in this post: the excess in the spike week over the trailing four week average, and the shortfall in the following fortnight against the same trend. The ratio between them is your show's pull-forward rate, and you will have it in about twenty minutes.